Removal of Viruses by Soil Passage: Overview of Modeling ... Microbe Removal/3 Virus Sp… · moved. Virus removal is the disappearance of viruses from the system and is defined in

Copyright © 2000, CRC Press LLC — Files may be downloaded for personal use only. Reproduction

of this material without the consent of the publisher is prohibited.

49

Critical Reviews in Environmental Science and Technology, 30(1):49–127 (2000)

Removal of Viruses by Soil Passage:Overview of Modeling, Processes, andParametersJack F. Schijven1 and S. Majid Hassanizadeh2

1Microbiological Laboratory for Health Protection, National Institute of Public Healthand the Environment, P.O. Box 1, 3720 BA Bilthoven, The Netherlands; 2DelftUniversity of Technology, Faculty of Civil Engineering and Geosciences, Delft, TheNetherlands

ABSTRACT: In this article, the modeling of subsurface virus transport under saturated conditionsand the factors that affect adsorption and inactivation are evaluated. Both equilibrium and kineticadsorption are considered. Equilibrium adsorption is found to be of little significance. Adsorptionappears to be mainly kinetically limited. At pH 7 and higher, conditions are generally unfavorablefor attachment, but viruses may preferentially attach to a minor surface fraction of soil grains that ispositively charged. The relation of pH with surface charge and their effects on sticking efficienciesare evaluated. Dissolved organic matter decreases virus attachment by competition for the samebinding sites and thus reduces attachment. Bonded organic matter may provide hydrophobic bindingsites for viruses and thus enhance attachment. Dissolved organic matter may disrupt hydrophobicbonds. The enhancing and attenuating effects of organic matter are very difficult to quantify and maybe responsible for considerable uncertainty when predicting virus removal.

Values of inactivation rate coefficients for attached viruses were calculated using data from somebatch studies. Enhanced or reduced inactivation is found to be virus-specific and almost independentof adsorption. Temperature is the most important factor that influences virus inactivation. Probablythe inactivation rate coefficients of free and attached viruses change similarly with temperature.

Some frequently used bacteriophages are evaluated as model viruses. MS2 and PRD1 meet therequirements for worst-case model viruses, at water temperatures less than about 10°C, at pH 6 to8, and if the soil does not contain too many hydrophobic sites and not too much multivalent cations.Bacteriophage ϕX174 may be a relatively conservative model virus, because of its low hydrophobic-ity and stability. Together in a cocktail, these three viruses span a range of properties, like size,surface charge, and hydrophobicity. F-specific RNA bacteriophages (FRNAPHs) may be very usefulnaturally occurring worst-case viruses. FRNAPHs that are present in surface water or treatedwastewater that is used for recharging groundwater, consist of stable and poorly adsorbing viruses.

An inventory of parameter values from field studies is made. Attachment appears to be the majorprocess that determines virus removal. Still, only very few data are available on attachment anddetachment of viruses under field conditions. Removal of viruses by soil passage, log(C/C0), appearsto decline nonlinearly with distance due to heterogeneities within the soil as well as within thepopulation of transported virus particles. Predictions of virus removal at larger distances are severelyoverestimated if they are based on removal data from column experiments or from short-distancefield studies.

KEY WORDS: virus adsorption, virus inactivation, virus transport models, virus removal, modelviruses, MS2, PRD1, ϕX174, F-specific RNA bacteriophages.



50

ABBREVIATIONS: dc [L], collector size; dp [L], virus particle size; foc, fraction of bonded organiccarbon; katt [T–1], attachment rate coefficient; kdet [T–1], detachment rate coefficient; keq [T–1],distribution coefficient; kF [L–2T], pseudo-first-order coefficient for the retarding effect of doublelayer repulsion on the adsorption rate; kirr [T–1], irreversible attachment rate coefficient; m, constantfor layering in Freundlich isotherms; n, soil porosity; v [LT–1], pore water velocity; t [T] time; x [L],distance; As, Happel’s porosity dependent parameter; C [L–3] number of free viruses per unit volumein the aqueous phase; C0 [L–3], initial number of free viruses per unit volume in the aqueous phase(t = 0); Ceq [L–3], number of free viruses per unit volume in the aqueous phase when at equilibriumwith attached viruses; D [L2T–1], hydrodynamic dispersion; DBM [L2T–1], diffusion coefficient;KB [J/K], Boltzmann constant; KL [L3], langmuir constant related to binding energy; Npe Pécletnumber; R, retardation or storage coefficient; RB, relative breakthrough; Seq [M–1], number per unitmass of soil for viruses attached to equilibrium sites; Skin [M–1], number per unit mass of soil forviruses attached to kinetic sites; Smax [M–1], maximum number per unit mass of soil for viruses whenall active surface sites are occupied; T [°C], temperature; U [LT–1], superficial water velocity; �,sticking efficiency; �L [L], longitudinal dispersivity; � collision efficiency; � [ML –1T–1], dynamicviscosity; �eff [T–1], inactivation rate coefficient for free viruses in suspension with soil; �l [T–1],inactivation rate coefficient for free viruses; �s [T–1], inactivation rate coefficient for attached viruses;�s,eq [T–1] inactivation rate coefficient for viruses attached to equilibrium sites; �s,kin [T–1] inactiva-tion rate coefficient for viruses attached to kinetic sites; �B [ML –3], bulk density of the saturated soil.

I. INTRODUCTION

A. Safe Drinking Water Production

Groundwater is the main source for drinking water production. Groundwatermay become contaminated with pathogenic microorganisms from artificial re-charge with wastewater or surface water, or from septic tanks or leaking sewagepipes. Therefore, to protect groundwater from contamination, adequate setbackdistances between these sources of contamination and production wells for drink-ing water are needed. Surface water is also a source for drinking water productionand is becoming increasingly important. Surface water may be contaminated withpathogenic microorganisms, mainly due to discharges of wastewater and by ma-nure run-off from agricultural land. To produce safe drinking water from surfacewater these pathogens need to be removed. One effective way is found to bepassage of surface water, through soil, as is the case in bank filtration, dunerecharge, and deep well injection. To assure production of safe drinking water fromsurface water, adequate travel times and travel distances are needed.

Pathogens of major threat to human health are viruses and the pathogenicprotozoa Cryptosporidium and Giardia. Little is known about the fate of thesepathogenic protozoa during soil passage (Hancock et al., 1998), although newinformation is emerging (Brush et al., 1999; Harter et al., 1999). Much moreinformation is available for viruses. It is believed that the processes that determineremoval of viruses during soil passage also apply to protozoa, albeit to a differentextent. Therefore, this review is confined to the study of virus removal by soilpassage. Viruses have been shown to be able to travel considerable distances



51

through the subsurface depending on their size, their adsorption characteristics,and their degree of inactivation (Keswick and Gerba, 1980). Nevertheless, soilpassage is considered an important barrier against viruses (Schijven and Rietveld,1996).

Currently, the extent of wellhead protection areas and bank filtration sites arebased on the travel time of the groundwater or recharge water. For example, inGermany (Dizer et al., 1984) and in The Netherlands a travel time of 50 to 60 daysis required. This is based on the assumption that a groundwater travel time of60 days is adequate to inactivate pathogenic microorganisms, to the degree that nohealth risk exists (Knorr, 1937). However, due to the high persistence ofCryptosporidium, Giardia, and viruses this may not be sufficient. Current knowl-edge of infection risks of, as a consequence of drinking water consumption, hasresulted in using maximum allowable concentrations for pathogenic microorgan-isms in drinking water. These are based on a maximum acceptable infection riskof one per 10,000 persons per year and dose-response relationships for pathogens(Regli et al., 1991). This approach has formed the basis for the Extended SurfaceWater Treatment Rule and it is under consideration for the Ground Water Disin-fection Rule in the U.S. (Macler, 1996). Based on the maximum level of infectionrisk, a proposal for drinking water protection policy is being prepared in TheNetherlands that leads to similar maximum allowable concentrations (Schijven etal., 1996). In the case of viruses, it is based on the dose-response relationship ofrotaviruses as a worst case. This maximum allowable concentration is 1.8 × 10–7

viruses per liter. Obviously, such a very low concentration is not directly measur-able. Therefore, the only way to evaluate the effectiveness of soil passage is tocalculate virus concentrations at the production point from the concentrations insource water by means of a computational model.

B. Modeling Transport and Fate of Viruses during Soil Passage

After a certain travel time and travel distance through soil, viruses are re-moved. Virus removal is the disappearance of viruses from the system and isdefined in this context as the logarithmic reduction of virus concentration,log10(C/C0). The processes of major importance for removal of viruses during soilpassage are adsorption and inactivation (Keswick and Gerba, 1980; Yates et al.,1987). Advection and dispersion affect spreading of viruses and thereby attenua-tion of virus concentrations. Modeling is a way to quantify these processes.Adsorption of viruses to soil may be modeled as either irreversible or reversible.In the case of irreversible attachment, there is no detachment. In the case ofreversible adsorption, one may have equilibrium and/or kinetic adsorption sites. Ingeneral, both kinds of adsorption may occur in a given medium. In this article,when talking of adsorption, the effects of both attachment and detachment aremeant. Thus, we consider a situation where viruses can adsorb to two different



52

kinds of sites on solid grains. There are sites where attachment and detachment arefast relative to the flow velocity, allowing equilibrium to occur. For some othersites, adsorption is kinetically limited relative to flow velocity, with constantattachment and detachment rate coefficients. The governing equations of solutetransport, including dispersion, advection, and inactivation for three-dimensionalsaturated flow are as follows:

nC

t

S

t

S

tn C n C QB eq B kin∂

∂∂ρ

∂∂ρ

∂+ + = ∇ ⋅ ⋅ ∇( ) ∇ ⋅ ( )D v– – (1)

S k Ceq eq= (2)

∂ρ∂

ρ ρB kinatt B kin s kin B kin

S

tnk C k S S= µ– –det , (3)

Q n C S Sl s eq B eq s kin B kin= µ + µ + µ, ,ρ ρ (4)

Here, C is the number of free viruses per unit volume in the aqueous phase, [L–3].In short, we refer to it as the free virus concentration. The adsorbed virus concen-tration is given in terms of number of viruses per unit mass of soil; we refer to itas the attached virus concentrations [M–1]. The symbols Seq and Skin are used todenote the concentrations of viruses attached to equilibrium and kinetic sites,respectively. Further, ρB is the bulk density of the saturated soil, [M.L–3] and n isthe porosity, [–]; D is the hydrodynamic dispersion tensor, [L2.T–1]; v is the porewater velocity vector, [L.T–1]; keq is a distribution coefficient, [L3.M–1]; katt and kdet

are the attachment and detachment rate coefficients, respectively [T–1]; µl is theinactivation rate coefficients for the free viruses, [T–1]; µl,eq and µs,kin are theinactivation rate coefficients for attached viruses to equilibrium and kinetic sites,respectively [T–1].

C. Studying Virus Adsorption and Inactivation at Batch, Column,and Field Scales

Adsorption of viruses to soil and concurrent inactivation can be studied atdifferent scales: in batch, column, and field experiments. In batch experiments,almost always only equilibrium adsorption is studied. Advection and dispersioncannot be investigated in batch experiments. In columns, transport of viruses isstudied often as a one-dimensional process. Columns can be made of packed soilmaterial or undisturbed soil. In the latter case, as in the field, effects of dispersionshould be considered. In uniformly packed columns, dispersion is usually smalland may be neglected. In field studies, the actual situation is investigated. Depend-ing on the hydrologic situation, transport may be modeled as 1-, 2-, or 3-dimen-sional. The effect of dispersion can be very important in the field.



53

In batch and column studies, any combination of soil and virus may beconsidered, whereas in field studies many restrictions apply. Removal of patho-genic viruses under field conditions can be studied only if contamination levels arehigh enough. Usually this is not the case. Only in exceptional situations, permis-sion may be obtained to seed pathogenic viruses in the field. At a site that is in usefor drinking water production, this will never be allowed. Therefore, model virusesthat are not pathogenic but are still representative for the transport behavior ofpathogenic viruses are needed. A model virus is suitable if its inactivation andadsorption are similar to that of pathogenic viruses under given conditions. Thisimplies that it should be possible to predict removal of pathogenic viruses bypassage through soil from the removal of the model virus.

Usually bacteriophages are used as model viruses. Bacteriophages offer thefollowing advantages:

• Bacteriophages are not pathogenic to human, but infect a specific hostbacterium.

• Bacteriophages can be prepared in large quantities (1010 to 1012 phages perml), allowing seeding of high numbers. This makes it possible to showremoval up to 11 log10.

• The assay of bacteriophages is relatively easy, whereas analysis of patho-genic viruses is much more complex, time consuming, and sometimes notpossible at all.

D. Purposes and Outline of the Article

Several reviews on the transport and fate of viruses through the subsurfacehave appeared. Keswick and Gerba (1980) reviewed reports on virus isolation fromgroundwater sources for drinking water production and from recharged groundwa-ter sites. Hydrogeological, biological, and meteorological factors affecting thesurvival and transport of viruses in groundwater were identified. The followingresearch needs were recommended: (1) investigation of virus-surface interactionsand virus inactivation in groundwater; (2) development of experimental methodsand predictive models; (3) development of criteria for adequate groundwaterprotection; and (4) performing field studies at land treatment sites, septic tanks, andon disease outbreaks by groundwater contamination. Gerba (1984) presented acomprehensive review on the factors influencing virus adsorption and inactivation,which were studied mostly in batch experiments. Yates et al. (1987) also reviewedthe factors affecting transport and inactivation of viruses through soils, includingmodeling of batch studies and of transport. They concluded that modeling of virustransport was constrained by a lack of quantitative information on virus behaviorduring transport. Yates and Yates (1992) presented an overview of the wayinactivation, equilibrium adsorption, advection and dispersion determine transport



54

of viruses. These factors were summarized in a quantitative manner by Gerba et al.(1992). Also of interest are the reviews on the transport of colloids in groundwaterby Swanton (1995) and Ryan and Elimelech (1996).

The main purposes of this article are to:

1. Evaluate the modeling that is used to describe and quantify adsorption andinactivation of viruses during subsurface transport;

2. Review the major factors that affect these processes;3. Evaluate model viruses as representatives of subsurface transport and behavior

of pathogenic viruses; and4. Discuss the relative contributions of adsorption and inactivation to the removal

of viruses during subsurface transport.

In Section II, modeling of equilibrium adsorption of viruses to soil in batchexperiments and during transport through soil in column and field experiments isdiscussed. A similar study but for kinetic adsorption will be presented in sectionIII. Also, in Section III, colloid filtration theory, Derjaguin-Landau-Verwey-Overbeek (DLVO) theory, hydrophobic interactions, and blocking are discussed.In Section IV, the factors that affect adsorption of viruses to soil are reviewed. InSection V, modeling of virus inactivation is reviewed, and the major factors thataffect inactivation are discussed in Section VI. This article focuses mainly onsaturated flow conditions; however, because unsaturated conditions have a signifi-cant impact on inactivation, this is discussed also. Section VII discusses the effectsof advection and dispersion on virus transport in the field. In Section VIII, somemodel viruses are evaluated. In Section IX, the relative contribution of adsorptionand inactivation to virus removal are evaluated, and removal of viruses withdistance is discussed. Finally, a summary and conclusions are presented in Section X.

II. EQUILIBRIUM ADSORPTION

A. Equilibrium Adsorption in Batch Experiments

In batch experiments, a suspension of viruses is agitated with a quantity of thesolid material of interest in a container. Concentration of viruses present in thewater phase of the container is measured as a function of time. Concentration ofviruses in a control container with water but without soil is measured to calculateinactivation, but also to be able to compensate for possible losses due to attachmentto the walls of the container. A typical semi-log plot of virus adsorption to soil ina batch suspension is given in Figure 1a. Initially, free virus concentrations declinewith time, but after a short time, they remain almost constant. At that point, adistribution of viruses between solid and liquid phase is obtained, because ofreversible adsorption. This apparent equilibrium is rapidly reached but is not



55

FIGURE 1. Adsorption of a virus to soil in a batch suspension: (a) Decreasing concentra-tion of free viruses with time; (b) Example of Langmuir isotherm.

instantaneous. It depends on the actual attachment and detachment rates. The timeto equilibrium has been reported to vary from 30 min (cf. Gerba and Lance, 1978;Goyal and Gerba, 1979; Taylor et al., 1980; Gerba et al., 1981; Singh et al., 1986;Gantzer et al., 1994) to 60 to 90 min (cf. Moore et al., 1981, 1982; Taylor et al.,1981; Bales et al., 1991; Sakoda et al., 1997). In all these studies, inactivation waseither neglected or found to be insignificant within the time scale of the experi-ment. At larger time scales, the free virus concentration continues to decrease ata steady rate due to inactivation.

Commonly in batch experiments, the kinetic behavior, which is operativebefore steady state is reached, is not considered, and values of attachment anddetachment rate coefficients are not determined. In batch studies, the apparentsteady-state concentrations are used to construct Langmuir or Freundlich iso-therms (Yates et al., 1987). The Langmuir model assumes that maximum attach-ment corresponds to a saturated monolayer of solute molecules on the adsorbentsurface, that the active sites for attachment are all the same, and that there is nointeraction between attached molecules. The Langmuir equation reads:

SS K C

K CeqL eq

L eq

=+

max

1 (5)

Where Seq is the concentration of adsorbed viruses and Ceq is the concentrationof free viruses after apparent equilibrium has been reached. Smax is the maximumadsorbed concentration when all active surface sites are occupied; KL is a constantrelated to the bonding energy. A typical example of a Langmuir isotherm is plottedin Figure 1b. It shows that the slope of the increasing part of the curve equals SmaxKL

and that the curve finally reaches Smax. When KLCeq << 1, the Langmuir equationmay be linearized to obtain a linear isotherm:



56

S S K Ceq L eq= max (6)

Freundlich isotherms have also been applied to describe attachment of viruses tosoil (Gerba, 1984). Here, no assumption is made on the homogeneity of active sitesfor attachment. The Freundlich formula reads:

S k Ceq eq eqm= (7)

Here, m is a constant. For many systems with low free virus concentrations, m isnot significantly different from unity, whereby Equation 7 reduces to a linearizedform similar to Equation 6 (Vilker and Burge, 1980; Yates et al., 1987).

Moore et al. (1981) showed that adsorption of poliovirus 2 to Ottawa sandcould be described by the Langmuir equation: Smax appeared to be 2.5 × 1012 virusparticles per kg of sand. At lower surface coverage, adsorption was successfullydescribed by the Freundlich equation. Vilker and Burge (1980) summarized sev-eral examples where Freundlich and Langmuir isotherms were applied. In theseexamples, SmaxKL was shown to vary between 2 and 640,000 liter/kg. In some ofthese examples, Smax was shown to be very large (i.e., in the order of 1014 to 1015

sites per kg of soil) but KL was shown to be very small in the order of 10–14 to10–11 liter per virus. Vilker and Burge (1980) concluded that virus adsorption issaturation limited, that is to say, the number of adsorption sites is finite. Theyfurther concluded that the large values of Smax and the small values for KL indicatethat virus adsorption is characterized by a large number of sites, but equilibriumstrongly favors the liquid phase over the adsorbed phase. Other examples ofapplication of Freundlich isotherms can be found in Gerba and Lance (1978),Taylor et al. (1980), Lipson and Stotzky (1983), Bales et al. (1991), and Sakodaet al. (1997).

In batch experiments, time to reach apparent equilibrium is not only dependenton the virus type and virus concentration, but it also depends on the particle sizeof the adsorbent and the degree of agitation (Vilker and Burge, 1980; Moore et al.,1981). Adsorption of poliovirus 2 was relatively constant from one experiment toanother, but adsorption to a few soils varied greatly. To organic muck 16 to 99%of the poliovirus was adsorbed, and to sandy loam 94 to 99.7% (Moore et al.,1981). High variability in adsorption among different batch studies is thought todepend primarily on the heterogeneity of soil preparations, like a wide range ofparticle sizes (Vilker and Burge, 1980). Jin et al. (1997) argued that results frombatch experiments have been neither consistent nor reproducible, largely due to thefact that there is no standard protocol. That is to say, different sizes and types ofcontainers and different methods of agitation are used. Also, an air–water interfacemay be present in some, but absent in other experiments. All these differences mayinfluence the equilibrium of a batch system. This makes it very difficult to comparevalues for adsorption between different batch studies. Nevertheless, batch tests



57

have been used extensively to investigate the effects of various factors (e.g., pH,organic matter, and soil type) and to compare adsorptive behavior of differentviruses in combination with different solid materials under a given set of experi-mental conditions.

B. Equilibrium Adsorption of Viruses to Soil during SubsurfaceTransport

Equilibrium adsorption has been assumed in several column and field studies,neglecting kinetic adsorption (e.g., Park et al., 1994; Powelson et al., 1990;Powelson and Gerba, 1994; Tim and Mostaghimi, 1991). Under these assumptionsand for a one-dimensional situation, Equations 1 and 4 are simplified to:

nC

t

S

tnD

C

xn

C

xQB

eq∂∂

ρ∂∂

∂∂

∂∂

+ =2

2 – – (8)

Q n C Sl s B eq= µ + µ ρ (9)

Equations 2 and 9 can be combined to:

RC

tD

C

xv

C

xQ n

∂∂

∂∂

∂∂

=2

2 – – / (10)

Here, the retardation coefficient is R = 1 + (ρB/n)keq, which is evidently equal toor larger than 1.

Viruses are not removed by equilibrium adsorption. According to the equilib-rium model description given by Equations 9 and 10, removal of viruses duringsubsurface transport is only due to virus inactivation. Sometimes an extra sink termfor irreversible attachment is also included to account for virus removal (Jin et al.,1997; Matthess et al., 1988; Yates and Ouyang, 1992):

RC

tD

C

xv

C

xk C Q nirr

∂∂

∂∂

∂∂

=2

2 – – – / (11)

Here, kirr is the irreversible attachment rate coefficient. It is usually denoted asfiltration (see Section III.D).

For solute contaminants and proteins, temporal and spatial variability of thedistribution coefficient has been observed, due to subsurface heterogenieties, suchas grain size, surface area, pH, temperature, and redox potential (Chrysikopoulosand Sim, 1996). Assuming the same may be the case with viruses, Chrysikopoulosand Sim (1996) developed a transport model with a stochastic time-dependent



58

distribution coefficient. Simulations with a time-dependent coefficient resulted inan enhanced spreading of the free virus concentration compared with the case witha constant keq.

In some cases, retardation coefficients of about 2 to 5 have been reported(Bales et al., 1991, 1997; Powelson et al., 1993; Powelson and Gerba, 1994).However, in most experiments, little or no retardation was found (Bales et al.,1991, 1993; Pieper et al., 1997; Jin et al., 1997; Schijven et al., 1999). Apparently,retarded breakthrough by equilibrium adsorption is of little significance.

Values of the “retardation coefficient” of less than one have also been reported.That is to say, faster virus transport relative to that of a conservative salt tracer hasbeen observed most probably due to pore size exclusion of the virus (see SectionVII).

III. KINETIC ADSORPTION

A. Kinetic Analysis of Batch Experiments

In a batch suspension of viruses and soil, adsorption equilibrium is not reachedinstantaneously. Instead, virus adsorption at the microscale can be described as theresult of two processes, each of which takes a certain time (see Gerba, 1984). Inthe first process, the viruses are transported close to the solid surface. Here, onespeaks of mass transport. In the second process, the viruses are immobilized at thesurface by physical and possibly chemical interactions. The overall rate of attach-ment depends on which of these two processes, mass transport or virus-surfaceinteractions, is the rate-limiting step (Grant et al., 1993). The kinetic behavior thatis operative before apparent equilibrium is reached can be described by virusattachment to the soil, virus detachment from the soil, and inactivation of free andattached viruses. Thus, the kinetics of the system can be characterized by fourparameters: katt, kdet, µl, and µs (Figure 2). The governing equations are

ndC

dtnk C k S n Catt B l= + µ– –detρ (12)

ρ ρ ρB att B s B

dS

dtnk C k S S= µ– –det (13)

As we have seen in Section II.A, on a time scale of a few hours, inactivation ofviruses in a batch system is commonly negligible. In that case, Equations 12 and13 have the following analytical solution:

C Ck k k k t

k katt att

att

/exp –det det

det0 =

+ +( )[ ]+ (14)



59

FIGURE 2. Attachment and inactivation of virus in bulk fluid and at the solid-liquid interface(Grant et al., 1993). Here, katt is the attachment rate coefficient; kdet is the detachment ratecoefficient; µl is the inactivation rate coefficient of viruses in the aqueous phase and µs thatof viruses attached to the solid surface.

The slope of the C–t curve at t = 0, assuming negligible inactivation, is approxi-mately equal to –katt (Figure 1a). This is evident from Taylor series expansion ofEquation 14, which yields:

C C k tt

k k kt

k katt att att att/ – –det det0

2 32

12 6

= + +( ) +( ) + ⋅⋅⋅ (15)

Thus, katt may be evaluated from early measurements of a batch experiment.Once an apparent steady state is reached (i.e., negligible dC/dt) and neglectinginactivation, from Equations 12 and 13 it also follows:

ρ ρBeq

atteq

Beq eq eqn

Sk

kC

nk C R C= = =( )

det

–1 (16)

Thus, from equilibrium results, kdet can be obtained. Therefore, it is possible todetermine both katt and kdet from batch experiments.

Often, adsorption of viruses to soil in batch experiments is expressed as thefraction of viruses that is adsorbed at equilibrium. The relation between the ratiokatt/kdet and the adsorbed fraction f, at equilibrium, is as follows:

k

k

C

C

f

fatt

eqdet

––

= =0 11 (17)



60

On longer time scales (e.g., days) virus inactivation will be significant and causesdisappearance of viruses from the system. The interplay of kinetic effects and virusinactivation in a batch system is discussed in Section V.C.

B. Kinetic Adsorption of Viruses to Soil during SubsurfaceTransport

Under transport conditions, adsorption of viruses to soil may be kineticallylimited relative to advection. In field studies (e.g., Bales et al., 1997; Schijven etal., 1999), it appeared that attachment rates were relatively fast compared withadvection, whereas detachment rates were much slower. Bales et al. (1997) ex-pressed this in terms of time scales for attachment or detachment relative toadvection over a characteristic travel distance L by v/(kattL) and v/(kdetL), respec-tively. It was found that the former term was about two orders of magnitude smallerthan one, whereas the latter was about three orders of magnitude greater than one.At very low flow rates (e.g., under some natural gradient conditions) kineticallylimited adsorption may successfully be described by equilibrium adsorption. Theattachment and detachment rates determine the shape of virus breakthrough curves.To show the differences and similarities between equilibrium and kineticallylimited adsorption, various breakthrough curves for a column were simulated usingthe CXTFIT-code (Toride et al., 1995). This code is based on analytical solutionsof equilibrium and kinetic transport models, including governing Equations 1 to 4for a one-dimensional situation. The breakthrough of virus seeded at a constantconcentration on top of a column for a duration of 10 days was calculated and isdisplayed in Figures 3a and 3b. The parameter values that were used for thesesimulations are given in Table 1. Dispersion is mainly determined by the charac-teristics of the porous medium, provided that no size and/or charge effects arepresent. A virus entering a soil therefore will experience the same dispersion as aconservative salt tracer. This is the case when considering curves A and B. Aconservative salt tracer will not be retarded and C/C0 reaches a plateau of oneshortly after breakthrough (curve A). A virus that does not attach to the soil willshow the same breakthrough as the salt tracer, but the plateau is lower due toinactivation of the virus (curve B). If a virus is also retarded due to equilibriumadsorption, it will show the same curve as B, but shifted to the right (curve C). Ifthis virus is subject to a higher dispersion, the shape of the curve becomes flatter,as illustrated by curve D. Here, a fraction of the virus breaks through earlier, butthe time to peak breakthrough is concurrent with the middle of the plateau ofcurve C.

Curve E simulates breakthrough of a virus, with the same dispersion as the salt,but now exhibiting kinetically limited adsorption, where the attachment and de-tachment rate coefficients are of the same order of magnitude. The time to peakbreakthrough is retarded. In fact, curve’s D and E in Figures 3a and 3b are verysimilar. Curve F simulates the virus, now exhibiting slow detachment. The shape



61

FIGURE 3. Simulated breakthrough curves of (A) a conservative salt tracer; (B) a virus thatdoes not adsorb, but that is inactivated; (C) a virus that is retarded due to equilibriumadsorption, and that is inactivated; (D) like C with higher dispersion; (E) a virus that adsorbsto kinetically limited sites, and that is inactivated; (F) like (E) but now detachment is muchslower; (G) like C with higher R; and (H) like E, but high katt and katt/kdet of 10. See Table 1for corresponding parameter values.



62

TA

BLE

1P

aram

eter

Val

ues

Use

d to

Sim

ulat

e B

reak

thro

ugh

Cur

ves

for

a C

onse

rvat

ive

Sal

t T

race

r an

d fo

r a

Viru

s E

xhib

iting

Diff

eren

t T

ypes

of

Ads

orpt

ive

Beh

avio

r

BC

DE

FH

AV

irus

Viru

s as

BV

irus

as C

Viru

s as

BV

irus

as E

GV

irus

as E

Sal

t–A

dsor

ptio

n+E

quili

briu

m+H

igh

+kin

etic

Slo

wV

irus

as C

Hig

her

k att,

Par

amet

erT

race

r+

Inac

tivat

ion

Ads

orpt

ion

Dis

pers

ion

Ads

orpt

ion

Det

achm

ent

Hig

her

Rk a

tt/k

det =

10

D0.

020.

020.

021

0.02

0.02

0.02

0.02

R1

13

3—

-11

—ka

tt—

——

—0.

750.

75—

10kd

et—

——

—0.

375

0.00

375

—1

µ l =

µs

00.

050.

050.

050.

050.

050.

050.

05

Not

e: R

esul

ting

brea

kthr

ough

cur

ves

are

show

n in

Fig

ures

3a

and

3b.

Por

e w

ater

vel

ocity

= 1

.5 m

/day

and

dis

tanc

e =

3 m

. D

imen

sion

of

D is

[m2 .

day–1

], th

at o

f k a

tt, k

det,

µ l a

nd µ

s is

[da

y–1].



63

of this curve on linear concentration scale is now the same as A and B; only themaximum concentration is lower due to attachment. However, when plotted on asemi-log scale a tail becomes visible, which makes it distinct from the other curves(Figure 3b). Now, if one stops measurements before the end of the plateau isreached, or if only linear plots are made, one may conclude that there is noreversible adsorption but a very high rate of irreversible adsorption (see Jin et al.,1997). Curve G simulates transport of a virus that is retarded due even more toequilibrium adsorption. Curve H shows that transport of a virus that attaches tokinetic sites may also show significant retardation. In this case the ratio of katt/kdet

is equal to R – 1 of curve G. Retardation is less in the case of H compared with G,but the breakthrough curve is much more dispersed.

These simulations show that description of virus transport by an equilibriummodel and a kinetic model may lead to similar results, and that investigation oftailing on a semi-log plot is needed to tell the difference. A shortcoming of manyvirus transport experiments is that the tail of the breakthrough curve is not mea-sured, so that the kinetic behavior cannot be observed (see Powelson et al., 1990;Powelson and Gerba, 1994; Jin et al., 1997).

Modeling of both equilibrium and kinetic adsorption was carried out by Baleset al. (1991, 1997). Adsorption of bacteriophages MS2 and PRD1 was shown to bereversible and kinetically limited (Bales et al., 1991, 1993, 1997; Kinoshita et al.,1993). The kinetic effect was evidenced from the slow rising limbs and the longtails of the breakthrough curves. Bales et al. (1991) showed that in order to fit anequilibrium model to MS2-breakthrough curves, apparent dispersion values of upto 150 times higher than for the salt tracer had to be used to fit the rising limbs ofthe breakthrough curve. Furthermore, to fit the declining limbs of the breakthroughcurve, a small dispersion value and an apparent R of less than one were needed.This is physically unacceptable and it indicates that the adsorption must be mod-eled as a kinetic process. Bales et al. (1991, 1997) found that the contribution ofequilibrium adsorption to the total adsorption was negligible, and that kineticadsorption controlled virus attenuation.

In the study of Bales et al. (1993), breakthrough curves of MS2 in silicashowed a gradual increase in free virus concentrations toward maximum break-through concentration, similar to curve E. This means that the detachment ratecoefficient is in the order of the attachment rate coefficient. However, the break-through curves of MS2 in columns with hydrophobic bonded silica reached asteady-state value much sooner, similar to curve F. In this case, detachment ismuch slower than attachment.

Other clear examples of kinetic behavior are shown for bacteriophages MS2,PRD1, ϕX174, Qβ and PM2 in a column study by Dowd et al. (1998), for MS2,PRD1, ϕX174, and poliovirus 1 in field studies by DeBorde et al. (1999), for MS2and ϕX174 by DeBorde et al. (1998); and for MS2 and PRD1 by Schijven et al.(1999). In these field studies, phage transport was not retarded by equilibriumadsorption. Figure 4 shows a typical breakthrough curve of MS2 from the field



64

FIGURE 4. Breakthrough curve of MS2 after 2.4 m of passage through dune sand(Schijven et al., 1999). Observed and simulated concentrations are shown. EC is theelectrical conductivity that was measured during breakthrough of the sodium chloride tracer.

study by Schijven et al. (1999). It was found that the value of katt determined mainlythe maximum breakthrough concentrations. It was also found that the tail slope wasdetermined mainly by the value of µs, whereas its intercept was affected mainly bythe value of kdet. The end of the rising limb and the start of the declining limb ofthe breakthrough curves could not be simulated completely, due to an as yetunknown process. Possibly, there is more than one type of kinetic sites involved.

There is growing evidence from column and field experiments that removal ofviruses during transport is governed mainly by kinetic adsorption. Neglect ofkinetic effects and use of equilibrium adsorption models may lead to an estimationof an artificially large dispersion coefficient. If detachment is slow compared withattachment, tailing may be observed only if measurements are carried out for a longenough time and when concentrations are plotted on a semi-log scale.

C. Batch vs. Column Experiments

Estimates of adsorption parameters from batch experiments appear to be oflimited use in the prediction of virus adsorption in column or field experiments. Asalready pointed out in previous sections, apparent equilibrium distribution iscommonly assumed in batch systems, whereas kinetic attachment/detachment isneeded to describe adsorption under transport conditions. Also, due to the stirringin a batch experiment, the number of accessible sites for attachment is much higherthan in a column. In a column, attachment rates may thus be so low that attachment



65

rates of various viruses are relatively close to each other. Detachment is limited bydiffusion over an energy barrier resulting from virus–soil interactions and bydiffusion across a boundary layer near the solid surface (Ryan and Elimelech,1996). The diffusion coefficient of the virus and the thickness of the boundarylayer control the rate of transport across the diffusion boundary layer. The latter iscontrolled primarily by the velocity of the advecting fluid. If this velocity in-creases, the thickness of the diffusion boundary layer decreases. Therefore, thedetachment rate coefficient in a batch system is presumably smaller than undertransport conditions, where there is advective flow. This implies that the observedratio katt/kdet from a batch experiment is expected to be larger than that obtainedfrom a column or field experiment.

Bales et al. (1991) found that adsorption of MS2 in columns with silica beadswas kinetically limited. They found ratio value a of 0.36 l/kg for katt/kdet in thecolumn experiment, but in the batch experiment katt/kdet was found to be 700 timeshigher. The 270 times higher specific surface area of the silica sorbent that wasused in the batch experiment may only partly explain this difference. Therefore,these findings suggest that katt/kdet from a batch experiment is indeed larger thanthat from a column or field experiment. Powelson and Gerba (1994) estimated theretardation coefficient R from column experiments with MS2 (R = 1.4), PRD1(R = 2.2), and poliovirus 1 (R = 5.2) using the equilibrium model as described byEquation 11. These estimates were 12 to 130 times smaller than those determinedfrom batch studies.

Contrary to expectations, in some batch experiments, values for the ratiokatt/kdet are found to be smaller than in corresponding column experiments. Goyaland Gerba (1979) carried out a number of batch experiments where the percentageof adsorption was determined after 30 min. Their results are converted to the ratiokatt/kdet using Equation 17, and are reported in Table 2. This ratio is generally foundto be less than one at pH 7 to 8. However, as discussed in the previous section, katt/kdet was found to be much larger than one in column studies (Bales et al., 1991,1993) and field studies (Bales et al., 1997; Pieper et al., 1997, DeBorde et al., 1999;Schijven et al., 1999). This is perhaps an artifact due to the duration of batchexperiments being too short. The following comparison between column experi-ments of Wang et al. (1981) and Lance et al. (1982) and the batch experiments ofGoyal and Gerba (1979) also seems to support this. Wang et al. (1981) and Lanceet al. (1982) studied the removal of poliovirus 1 and echoviruses 1 and 29 incolumns with the same soils. Residence time of the viruses in the columns wasabout 3 days. It was found that poliovirus 1 and both echoviruses were all removedvery well (99 to 99.9%). Lance et al. (1982) suggested that poliovirus 1 does notbehave significantly different from other enteroviruses under transport conditionsin long soil columns that approximate field conditions. However, in the batchexperiments of Goyal and Gerba (1979), poliovirus 1 adsorbed very well todifferent soils, whereas echovirus 1 and 29 were the least adsorbing viruses(Table 2). Thus, the difference in adsorption that was found in batch experiments



66

TA

BLE

2V

alue

s of

the

Rat

io

k att/k

det f

rom

Bat

ch E

xper

imen

ts w

ith V

iruse

s fr

om G

roup

I,

II, a

nd I

II an

d N

ine

Diff

eren

t S

oils

pH4.

54.

95.

57.

17.

17.

88

88.

2pI

a

IC

B4

(V21

6)14

22.

00.

790.

140.

020

0.43

0.52

0.47

0.52

CB

4 (V

240)

7612

2.7

1.3

00

0.32

0.35

0.54

Ech

o 1

(Far

ouk)

332

93.

50.

390.

271.

20.

140.

120.

155.

1E

cho

1 (V

212)

332

2.3

3.8

0.79

0.30

0.85

0.61

0.23

0.01

06.

4E

cho

1 (V

239)

491.

92.

10.

053

0.11

00.

961.

11.

15.

3E

cho

1 (V

248)

499

999

4.6

00.

330.

430.

330.

490.

495.

0ϕX

174

4999

324.

90.

220.

0050

00.

140

06.

7M

S2

999

0.89

0.52

2.2

0.20

0.20

1.3

0.64

0.43

3.5b

Ave

rage

929

132

2.9

0.64

0.15

0.39

0.54

0.43

0.40

IIC

B3

(Nan

cy)

9.0

4.9

2.7

1.3

1224

3232

0.54

Ech

o 7

(Wal

lace

)99

2.1

0.08

70.

2549

999

199

990.

28P

olio

1 (

LSc)

999

161.

30.

7219

999

249

994.

66.

6T

299

9924

91.

80.

087

9.0

0.56

196.

12.

34.

2T

433

233

26.

10

492.

612

433

216

4–5

Ave

rage

2288

121

2.4

0.47

2840

512

511

44.

7III

f299

0.06

40

00.

350

00.

190.

0030

Soi

l Cha

ract

eris

tics

f oc

%0.

31.

40.

43.

64.

20.

880.

781.

70.

27C

lay

(%)

3953

43

543

3628

13S

ilt (

%)

1316

48

208

2413

10S

and

(%)

4813

9289

2689

4059

77

apI

-val

ues

take

n fr

om G

erba

(19

84).

bpI

for

MS

2 fr

om P

enro

d et

al.

(199

5).

Not

e: D

ata

deriv

ed f

rom

Goy

al a

nd G

erba

(19

79).



67

was not found in column experiments. An explanation for this difference may bethat echoviruses attach much slower than poliovirus 1, and thus may not havereached equilibrium after 30 min in the batch tests, whereas in the columns theseviruses had sufficient time to attach to a similar extent. The following analysisfrom a field study by Schijven et al. (1999) also illustrates that a much longer timemay be needed to reach equilibrium in a batch experiment. In this study, a valuefor katt/kdet of about 5000 was found (see also Table 8). Using Equation 14 and thevalues from this field study, it can be shown that apparent equilibrium would bereached after 40 h in a batch experiment. Now, if one stops this batch experimentafter, for instance, only 1 h, a ratio katt/kdet of only 0.18 would be found. Thus, itis believed that in many cases, batch experiments may have been stopped far toosoon, thereby causing a strong underestimation of katt/kdet. Also, this would meanthat a low value of katt/kdet that is found in a batch experiment after a short periodof time is rather a consequence of slow adsorption than of equilibrium.

Nevertheless, it may not always be the case that katt in batch experiments islarger than in column experiments. An example is reported in Jin et al. (1997).They stopped a batch experiment of ϕX174 and sand after 3 h and estimated aretardation coefficient R of 4.5. This means they assumed equilibrium adsorptionwas reached. Then, based on this value, they predicted a retarded breakthroughcurve for their column experiments. However, breakthrough of ϕX174 was notfound to be retarded in any of the column experiments (R = 1), implying that theremoval process was not due to equilibrium adsorption. As already pointed out inthe previous section, measurements in the column experiment were stopped beforethe end of the plateau of the breakthrough curve was reached. Detachment in thecolumn experiments therefore was not investigated. Thus, the authors interpretedthe removal process as a first-order irreversible attachment in both column andbatch experiments. This interpretation then indicated that katt of ϕX174 obtainedfrom the column experiment was found to be about twice as high as that from thebatch experiment. At this point there is no clear explanation for these observations.

D. Colloid Filtration

Analogous to attachment of viruses to solid surfaces in a batch system, attach-ment of viruses in flowing water to the surfaces of solid particles in a porousmedium involves two processes: mass transport to the surface, and virus-surfaceinteractions (see Section III.A). Therefore, the attachment rate coefficient katt

depends on microscale flow and diffusion characteristics as well as surface prop-erties of viruses and soil grains. These processes are also described by colloidfiltration theory, which allows exclusion of the effects of flow and diffusion byexpressing the attachment rate of viruses in terms of collision efficiency η andsticking efficiency α. According to this theory, a suspended particle may come intocontact with a particle of the solid medium, the collector, either by interception,



68

sedimentation, or diffusion (Yao et al., 1971). The attachment rate coefficient isrelated to the collision efficiency η and the sticking efficiency α as follows (Yaoet al., 1971):

kn

dvatt

c

= ( )32

1– αη (18)

Here, dc is the average diameter of collision (grain size), [L]. The fraction ofparticles that collide with the collector is given by η, the collision efficiency.Viruses can be regarded as colloidal particles because of their size and surfacecharge. Viruses are small in size and their transport in the immediate vicinity of thecollector surface is dominated by Brownian diffusion, whereas effects of intercep-tion and gravitation are negligible. In this case the collision efficiency is given bythe Smoluchowski-Levich approximation (Penrod et al., 1996):

η = 4 1 3 2 3A Ns Pe/ – / (19)

Here, NPe = dcnv/DBM, a Péclet number, accounts for diffusion; DBM = KB(T + 273)/(3πdpµ) is the diffusion coefficient, [L2T–1]; KB = 1.38 × 10–23 is the Boltzmannconstant [J/K]; T is temperature; dp is the virus particle size; µ is the dynamicviscosity [ML–1T–1]; As = 2(1 – γ5)/(2 – 3γ + 3γ5 – 2γ6) is Happel’s porositydependent parameter, with γ = (1 – n)1/3.

Funderburg et al. (1981), Wang et al. (1981), and Lance et al. (1982) found thatvirus removal in columns was inversely related to the flow velocity. Data onremoval of poliovirus 1 and echovirus 1 in the upper 17 cm of the columns in thestudy by Wang et al. (1981) offer the possibility to investigate the relation betweenvirus removal and flow rate. In a column with characteristic length L, assumingsteady-state conditions, and neglecting dispersion, virus inactivation, and detach-ment, the solution of Equation 1 is:

log –.

–.

–– / /

/C

C

k

v

Lv L

n

dA

D

d natt

cs

BM

c0

2 3 1 3

2 3

2 3 2 332

14

= = = ( )

ψ ψ αwhere (20)

Thus, colloid filtration theory predicts that virus removal, log(C/C0), is propor-tional to v–2/3 (Yao et al., 1971). In the study by Wang et al. (1981), it was foundthat poliovirus 1 and echovirus 1 were removed to a similar extent at different flowrates. Figure 5 shows removal of both viruses vs. v.–2/3 This relation appears to belinear with a correlation coefficient of 90%. Also, Jin et al. (1997) observed lessremoval of ϕX174 when flow velocity was increased. The ratio of virus removal



69

FIGURE 5. Removal, log(C/C0) of poliovirus 1 and echovirus 1 vs. v–2/3 (data from Wanget al., 1981).

at the different flow velocities was approximately equal to the ratio of the two flowvelocities to the power of –2/3. This suggests that the relation between virusremoval and flow velocity, as described by colloid filtration theory, is valid.

The sticking efficiency, α, represents the fraction of the particles colliding withthe solid grains that remain attached to the collector (Martin et al., 1992). Thesticking efficiency reflects the net effect of repulsive and attractive forces betweenthe surfaces of the particles and the collector and depends on the surface charac-teristics of the virus and soil particles. Therefore, it depends on pH, organic carboncontent, and ionic strength. It is believed that α is independent of hydrodynamiceffects (velocity and dispersion). So, if for a given set of conditions such as typeof virus, type of soil, pH, organic carbon content, and ionic strength α is known,it is possible to calculate the value of katt for a different set of physical conditionsusing Equation 18. This emphasizes the importance of knowing the value of αunder a given set of conditions.

Commonly, α is derived from experimental values of katt and calculated valuesof the collision efficiency η using Equation 18 (Bales et al., 1991, 1993; Kinoshitaet al., 1993; Penrod et al., 1996; Redman et al., 1997; Schijven et al., 1998, 1999).Alternatively, α has been estimated from relative breakthrough (RB) of the massof viruses relative to that of a conservative salt tracer (Pieper et al., 1997; DeBordeet al., 1999; Ryan et al., 1999) and assuming irreversible adsorption using thefollowing equations:



70

αα

ηα=

( ) ( )[ ][ ]( )

=

∫ ∫

d x RB

nRB

C

Cdt

C

Cdt

c L

L t

tvirus

virus t

tsalt

salt

f f1 2 1

6 1

2

0 00 0

– / ln –

–

, ,

where

(21)

The sticking efficiency can also be calculated according to the so-called InteractionForce Boundary Layer (IFBL) approximation (Swanton, 1995; Ryan and Elimelech,1996):

α ββ

β β=+

( ) ( )

1

213

1 3 1 3

1 3

S AD

Ud

k d

DsBM

c

F c

BM

where =13

– / – /

/

Γ (22)

Here, kF is a pseudo-first-order coefficient that accounts for the retarding effect ofdouble layer repulsion on the attachment rate, [L–2T]. S(β) is a slowly varyingfunction of β with tabulated numerical values, and U is the superficial flowvelocity, [LT–1] (Ryan and Elimelech, 1996). As described in the next section,theoretical values of the sticking efficiency, and thus also of the attachment ratecoefficient, considerably underestimate experimental values.

E. Derjaguin-Landau-Verwey-Overbeek Theory

The attractive and repulsive forces that exist between colloids (e.g., viruses)and grain surfaces determine the interactions between them. These forces can bedescribed in profiles of the intersurface potential energy by DLVO theory. Suchprofiles are constructed by summing the potential energies of double-layer repul-sion or attraction, London–van der Waals attraction, and poorly characterizedshort-range “non-DLVO” forces such as hydration and steric repulsion (Ryan andElimelech, 1996). An example of a DLVO-energy profile is shown in Figure 6. Thetotal potential energy curve is characterized by an attractive energy well at a verysmall separation distance (δ), with the primary minimum (φmin1), a repulsive energybarrier (φmax), and a shallow attractive energy well at a larger separation distance,with the secondary minimum (φmin2).

The van der Waals attraction is an electrical force between instantaneousdipole moments within the different molecules (Gerba, 1984). It is linearly depen-dent on the value of the Hamaker constant. The Hamaker constant depends on thenature of the interacting materials (Swanton, 1995). The Hamaker constants ofmost forms of organic matter are similar to those of water, hence the van der Waalsinteractions between organic colloids are weak. Inorganic matter tends to havelarge Hamaker constants and thus a stronger inherent attraction for virus (Mooreet al., 1981).



71

FIGURE 6. DLVO energy as a function of separation distance between a colloid and acollector. The total potential energy (φTotal) is the sum of the double layer potential energy(φDL), the van der Waals potential energy (φvdW), and the Born potential energy (φBorn). Thetotal potential energy curve is characterized by an attractive well at a very small separationdistance (φ), the primary minimum (φmin1), a repulsive energy barrier (φmax), and a shallowattractive well at a larger separation distance (φmin2). The potential energy is normalized bykBT (Ryan and Elimelech, 1996).

A double-layer potential energy arises from the overlap of diffuse clouds ofions (double layers) that accumulate near charged surfaces to balance the surfacecharge. If the interacting surfaces are like-charged, the double-layer potentialenergy will be repulsive (Ryan and Elimelech, 1996). If the surfaces are oppositelycharged, the double-layer potential energy will be attractive. In formulations of thedouble layer theory, potential energy is considered to be sensitive to variations inthe surface potentials of the colloid and the collector, ionic strength of the solution,and colloid size. The dependence on colloid size has not been verified (Ryan andElimelech, 1996).

It has been shown that the rate coefficients for attachment (katt) and detachment(kdet) should increase exponentially as the height of the corresponding energybarriers increases (Ryan and Elimelech, 1996):

kk Tatt

B

∝

exp – maxφ

(23)

kk TB

detmax minexp –

–∝

φ φ 1

(24)



72

DLVO theory provides a conceptual framework to understand interactions of virusparticles and solid surfaces under different conditions, such as pH, ionic strength,and colloid size. The effects of these conditions are discussed in more detail inSection IV.

DLVO theory has, however, several shortcomings in predicting attachmentand detachment rates. It has generally been shown that under unfavorable condi-tions for attachment, DLVO theory underestimates attachment by many orders ofmagnitude. At pH 7 to 8, as in many aquifers, the net surface charge of most virusesand soils is negative, and thus conditions for attachment are generally unfavorable.Under such conditions, colloid–surface interactions are the rate-limiting processfor attachment (Ryan and Elimelech, 1996). Experimentally determined stickingefficiencies are sensitive to the ionic strength of the solution and to the electroki-netic (zeta) potentials of particles and collectors but not to the degree predicted bytheory. Experimentally determined sticking efficiencies and attachment are virtu-ally independent of colloid particle size, in contrast with DLVO theory predictions.Also, it has been shown that the sensitivity of attachment rates to ionic strengthdecreases markedly as the degree of surface charge heterogeneity increases (Ryanand Elimelech, 1996). Furthermore, DLVO theory was formulated for smoothbodies with ideal geometries (e.g., spheres interacting with flat surfaces) anduniform properties. In practice, real particles are irregular, and the surface of thebodies are rough and are likely to be heterogeneous in composition and charge(Swanton, 1995).

Under favorable conditions (i.e., in the absence of repulsive energy barriers orin the presence of attractive double-layer interactions) colloidal transport to thevicinity of the soil surfaces is the rate-limiting step. In this case, particle attachmentmodels within the framework of the DLVO theory are satisfactory in predicting theeffect of solution ionic strength, fluid velocity and particle size. Favorable chemi-cal conditions for attachment may develop in groundwaters with high levels ofwater hardness and ionic strength. Attachment will also be favorable for solidsurfaces (or patches on solid surfaces) that are positively charged due to iron,aluminium, or manganese oxide coatings (Ryan and Elimelech, 1996). Because ofthe electrostatic attractive forces, it is reasonable to assume that attachment to suchfavorable patches is irreversible (Ryan and Elimelech, 1996).

Bales et al. (1991) suggested that attachment to kinetically limited sites wouldoccur in the primary minimum, and to fast equilibrium sites in the secondaryminimum. However, as was shown in their and other studies (Bales et al., 1993,1997; Kinoshita et al., 1993; Schijven et al., 1999), kinetically limited attachmentprevails. Loveland et al. (1996) calculated DLVO profiles for PRD1 and quartzwith or without ferric oxyhydroxide coatings and showed that secondary minimawere extremely small. So, probably, attachment of a virus in the secondary mini-mum of a site is not significant. Loveland et al. (1996) suggested that a secondaryminimum does need to be deep to affect virus transport. Under strongly repulsiveconditions, viruses will not attach in the primary minimum. Instead, viruses may



73

accumulate in the boundary layer in front of the energy barrier. In time, diffusionand advection will carry them out of the boundary layer and back to the advectiveflow. Regardless of the depth of the secondary minimum, the transport of theseviruses has been retarded by reversible attachment.

F. Hydrophobic Interactions

As reviewed by Gerba (1984), hydrophobic interactions between viruses andsolid surfaces may also contribute significantly to adsorption. Hydrophobic inter-actions may be seen as a consequence of the thermodynamically unfavorableinteraction of hydrophobic substances with water molecules and is not due tointeractions among hydrophobic particles themselves (Wait and Sobsey, 1983).Hydrophobic interactions are not described by DLVO theory (Swanton, 1995).Interactions between hydrophobic groups on the surfaces of the virus and the solidmay cause an increase in virus attachment. At high pH, when both the surfaces arenegatively charged, hydrophobic interactions may be the major factor maintainingvirus attachment (Shields and Farrah, 1983; Gerba, 1984; Bales et al., 1991).Bales et al. (1991) showed that MS2 attached much less to silica beads in abatch experiment at pH 7 than at pH 5, but when the silica beads were coated withC18-trichlorosilane, 400 times more attachment took place, independent of pH.Bales et al. (1991) concluded that hydrophobic effects are important for adsorptionof even relatively hydrophilic viruses. Loveland et al. (1996) argued that theaddition of hydrocarbon chains to a silica surface decreases the negative surfacecharge of the silica surface in addition to providing hydrophobic attachment sitesfor viruses. According to Loveland et al. (1996), increased virus attachment may,therefore, be more reasonably attributed to a decreased double-layer repulsionrather than to hydrophobic expulsion from solution.

Hydrophobic interactions of viruses with solid surfaces have been suggestedto play a role in the interaction between poliovirus 1 and millipore and zeta plusfilters (Farrah et al., 1981), and also, among poliovirus 1, echovirus 1, and rotavirusSA 11 and highly organic estuarine sediments (Wait and Sobsey, 1983) andbetween poliovirus 1 and nitrocellulose membrane filters (Shields and Farrah,1983). Bales et al. (1991, 1993) suggested hydrophobic interactions of MS2 andPRD1 with octadecyltrichlorosilane coated silica. It was also suggested for MS2,ϕX174, T7, PRD1, and ϕ6 with nitrocellulose and cationic polysulfone membranes(Lytle and Routson, 1995).

G. Blocking

When particles attach to a solid surface, the attachment rate may decrease withtime if particle–particle interactions are repulsive; the solid surface becomes



74

progressively occluded as particles accumulate. This surface exclusion phenom-enon is termed blocking. Ryan and Elimelech (1996) have reviewed the blockingprocess of colloids and how it is modeled. Blocking may prevail in groundwaterhaving low ionic strength and low levels of hardness. Because the majority of theavailable surface area of solids in groundwater has chemical characteristics unfa-vorable for particle deposition, colloidal attachment in groundwater is thought tobe largely restricted to a minor patch-wise distributed fraction having energeticallyfavorable charge characteristics (Ryan and Elimelech, 1996). The same restrictionsapply to attachment of viruses. Jin et al. (1997) observed blocking of ϕX174 in asand column. Initially, significant removal of ϕX174 took place, but with increas-ing amounts of attached phages, the attachment rate dropped and an increase ofC/C0 up to one was observed. However, under field conditions, virus concentra-tions in recharging water are so low that blocking of binding sites with viruses isnot likely to occur. In case of contamination of an aquifer with wastewater, virusesrepresent only a very minor fraction of the wastewater organic matter. In that case,organic matter will probably block binding sites and a progressive blocking effectof viruses will not be the case.

IV. FACTORS AFFECTING ADSORPTION OF VIRUSES TO SOIL

A. Introduction

The interactions that take place between viruses and soil particles are deter-mined by their surface characteristics. Virus–soil interactions are electrostatic andhydrophobic in nature. Surface characteristics may be altered by changes in pH,ionic strength, multivalent ions, and organic matter (Gerba, 1984). Changes inthese interactions are quantified by changes in attachment and detachment ratecoefficients. As we have seen in Section III.E, DLVO theory provides a conceptualframework to understand many of these changes. In the following sections, theeffect of virus and soil type, pH, ionic strength, multivalent cations, and organicmatter on attachment and detachment of viruses to solid surfaces is discussed inmore detail.

B. Effect of Virus Type

The behavior of different viruses in their interactions with solids is believed tobe the result of differences in the electrical charge and the hydrophobicity of thevirus surface (Shields and Farrah, 1987). Most viruses have a size in the range of20 to 200 nm and consist of nucleic acid encapsulated in a protein capsid. Theamino acids in this protein coat contain weakly acidic and basic groups, like



75

carboxyl and amino groups, that determine the amphoteric nature of the virusparticle. The isoelectric point (pI) of a virus particle is the pH value at which it hasa zero net charge. However, localized pockets of positive and negative chargesmay still exist across the virus surface. The isoelectric point of a virus may varynot only by the type of the virus but also by the strain (Gerba, 1984). At pH valuesabove 5, the electrophoretic mobility of MS2, a measure of its surface charge,remains constant (Penrod et al., 1995). At pH values above 6, electrophoreticmobilities of vaccinia virus, reovirus and λ are also relatively insensitive tochanges in pH (Penrod et al., 1995). At higher pHs, PRD1 (Loveland et al., 1996)and recombinant Norwalk-like virus particles (Redman et al., 1997), however,show a further decrease in their negative charge.

The coat proteins of a virus may contain spans of amino acids that arehydrophobic. Dependent on the way these proteins are folded, such hydrophobicparts may either be on the inside or the outside of the virus coat (Gerba, 1984).Viruses differ in hydrophobicity, as was shown by Shields and Farrah (1987), whodetermined the hydrophobic nature of 15 viruses by means of octyl-sepharosechromatography. Echovirus 5 and MS2 were found to be the most hydrophobic ofthe viruses tested, whereas echovirus 7 and ϕX174 were found to exhibit little ifany hydrophobic character. According to Lytle and Routson (1995) in a study onthe retention of viruses by nitrocellulose and cationic polysulfone membranes,ϕX174 was the least hydrophobic of the viruses studied, but hydrophobicity ofMS2, T7, PRD1, and ϕ6 was not much different. A detailed description of the coatproteins of MS2, including its hydrophobic outer regions, is presented by Penrodet al. (1996).

Adsorption behavior of viruses is very complex, and even under relativelywell-defined conditions of batch experiments a nonunique behavior is observed.Several studies (Burge and Enkiri, 1978; Goyal and Gerba, 1979; Singh et al.,1986) have clearly shown that most viruses, even strains of the same type, maybehave differently under similar conditions. Goyal and Gerba (1979) comparedattachment of a large number of different types and strains of human enteroviruses,bacteriophages, and a simian rotavirus to nine different soil types. From these data,it was possible to distinguish two groups of viruses according to their meanpercentage of adsorption at equilibrium (Gerba et al., 1981). It was remarked thatthis grouping depended on the choice of viruses and soils that were studied.Viruses of group II adsorb very well to most soils, sludges, and marine sediments,whereas those of group I adsorb to a lesser degree. Mean percentage of adsorptionof viruses of group I was 44% and that of group II 78%. Bacteriophage f2 isprobably a special case because its mean adsorption of 16% was significantlylower than that of all the other viruses. Table 2 summarizes the batch experimentsof Goyal and Gerba (1979). Recall that instead of giving percentages of adsorptionat equilibrium, the ratio katt/kdet is calculated using Equation 17. Adsorption ofgroup I viruses is most affected by pH, exchangeable iron content and organic



76

carbon content of the soil. Adsorption is inversely correlated with pH, whichexplained 83% of the variation in the mean attachment. Adsorption of group Iviruses is also inversely correlated with the organic matter content of the soil, butpositively correlated with exchangeable iron. Adsorption of group II viruses wasnot found to be significantly affected over the range of soil characteristics studiedhere. Gerba et al. (1981) suggested that the net charge of group I viruses wasinfluenced more over the range of pH values and organic matter concentrationsfound in nature than that of group II viruses.

Several studies indicate that knowledge of the pI of a virus makes it possibleto predict the likelihood of its attachment to a charged surface (Gerba, 1984). Also,Gerba (1984) suggested that group I viruses might have lower isoelectric points(pI) than group II viruses. However, the data in Table 2 are not in agreement withthat suggestion. Dowd et al. (1998) observed a linear relationship between theamount of adsorption to soil and the isoelectric point of five different phages, MS2,PRD1, Qβ, ϕX174 and PM2 in circulating flow columns (i.e., in these columns theoutflow is connected to the inflow). The amount of adsorption was the least for thephages with the higher pI. This is contradictory to the expectation that a virus withhigher pI is less electronegatively charged and thus adsorbs better because ofweaker repulsion. Besides, in flow-through columns this relation with pI was notfound. Possibly, the presence of multivalent cations was the cause that morenegatively charged viruses attach more.

The isoelectric point is only an indication of what the net surface charge of avirus may be at a given pH. Penrod et al. (1996) and Redman et al. (1997) showedthat λ and Norwalk virus-like particles have a stronger negative zeta-potential thanMS2 at pH 5 to 7, although MS2 has a lower pI than the other two. Penrod et al.(1996) found strong evidence that the surface charge of MS2 originates from thecharge of the amino acid residues that are present on the exterior of the virusparticle, by calculating pI of MS2 correctly from the ionization of these exteriorlylocated amino acid residues. They also suggested that the attachment rate of MS2is not only determined by electrostatic repulsion, but also by steric repulsion ofhydrophilic loops that protrude from the surface of MS2. In this study, it was alsoshown that the attachment rate of phage λ was determined primarily by the surfacecharge on the capsid of this virus, but the tail of λ contributes relatively little to theoverall charge of the virus.

Due to the differences in electrical charge and hydrophobicity that existbetween different types of viruses, and even between different strains of the samevirus type, virus–soil interactions will differ, and thus virus removal by soilpassage will be virus-type dependent. Therefore, to predict removal of viruses bysoil passage, a combination of viruses may be considered that represent a range ofadsorption characteristics. Alternatively, a virus that adsorbs less than other virusesunder certain conditions, may be considered as a worst-case model virus. Viruseswith a strong negative surface charge and little hydrophobicity meet these require-ments.



77

C. Effect of Soil Type

Ryan and Elimelech (1996) described that most of the surface area of soilgrains has chemical characteristics unfavorable for particle deposition. Therefore,colloidal attachment in groundwater is thought to be largely restricted to a minorfraction of the grain surface having energetically favorable charge characteristics.These favorable sites, resulting from surface charge heterogeneity, are often mani-fested as positively charged patches on the surface of negatively charged mineralgrains. Such patch-wise charge heterogeneities are general to all aqueous geologicsettings, originating from inherent differences in the surface properties of adjacentcrystal faces on mineral grains, and from minerals having bulk- or surface-boundchemical impurities. Oxides of iron, aluminium, and manganese are the mostcommon sources of surface charge heterogeneity in the groundwater environment.These oxides carry a positive charge at near-neutral pH and are generally presentin minor amounts as surface coatings on mineral grains. It has been shown thatminor degrees of charge heterogeneity on collector surfaces result in attachmentrates that are orders of magnitude larger than similar surfaces having no chargeheterogeneity.

Several studies appear to support this concept that viruses preferably attach toa surface fraction of the soil having favorable charge characteristics. Solids thathave high isoelectric points are better virus adsorbents than those with low isoelec-tric points (Gerba, 1984). Also, favorable for attachment of viruses are a highercation exchange capacity (Burge and Enkiri, 1978), exchangeable iron (Gerba etal., 1981), and iron oxides (Moore et al., 1981; Lipson and Stotzky, 1983; Grantet al., 1993; Loveland et al., 1996). Attachment of viruses has been shown to bebetter to soils with higher specific surface areas (Moore et al., 1982). Generally,granular soils are weaker adsorbents than clays and minerals (Sobsey et al., 1980;Moore et al., 1981). Clays may have surfaces that have a very heterogeneouscharge distribution. Vilker et al. (1983) suggested that poliovirus 1 adsorbed to theedges of montmorillonite particles, where regions of positive charges are locateddue to the presence of aluminium ions. Farrah and Preston (1993) modified sandby precipitation of metallic salts. Columns of this modified sand removed signifi-cantly more viruses (poliovirus 1, coxsackievirus B5, echovirus 5, and MS2) thancolumns of unmodified sand. In batch experiments, Moore et al. (1981) showed anapparent surface saturation of poliovirus 2 (about 30 nm in size) to Ottawa sand atpH 7.5 of 2.5 × 1012 virus particles/kg. It was thus concluded that granular materialslike Ottawa sand (surface area 0.018 m2/g) have an enormous capacity for virusesand are unlikely to become saturated in natural systems. Nevertheless, at thismaximum concentration of attached virus particles, only approximately 1% of thetotal sand surface was covered. Jin et al. (1997) demonstrated blocking of ϕX174in columns with Ottawa sand. Ottawa sand contains detectable levels of iron,probably in the form of goethite. By calculation of the surface area that is occupiedby a monolayer of ϕX174 particles, they also showed that the apparent saturation



78

of the sorption sites was far less than the total surface area of the sand. Therefore,they suggested that the active surface for virus attachment was limited to patchesof positive charges on the sand particles formed by goethite.

D. Effect of pH

In many batch studies, it has been shown that, generally, viruses attach to alesser extent at higher pH (Burge and Enkiri, 1978; Goyal and Gerba, 1979; Sobseyet al., 1980; Taylor et al., 1980, 1981; Gerba et al., 1981; Grant et al., 1993).According to DVLO theory, this can be explained by an increased electrostaticrepulsion at higher pH. Figure 7 shows a plot of the average values of the ratio katt/kdet vs. pH for viruses of group I and II from the study of Goyal and Gerba (1979).In the range of pH 4.5 to 5.5 there is little difference between viruses of group Iand II. Goyal and Gerba remarked that the naturally isolated viruses used in theirstudy were all isolated from field samples by adsorption to membranes at low pH.Thus, viruses that did not have this characteristic would not have been isolated.Within the pH range of 4.5 to 5.5, the ratio katt/kdet declines with increasing pH. Theratio for group I viruses continues to decrease, but at a lower rate, as pH increases.It appears that the ratio katt/kdet of group I viruses becomes less than 1 at a pH equalto or higher than their pI. However, the ratio katt/kdet of a group II virus shows anincrease for pH values of up to 8, at which point it also decreases with increasingpH (see also Table 2).

Tables 3 and 4 are an inventory of sticking efficiencies of viruses in columnand field studies, at different pH values, respectively. In Figure 8, the sticking

FIGURE 7. Average values of ratio katt/kdet vs. pH of viruses of group I and II to soil in abatch suspensions (Goyal and Gerba, 1979). See also values in Table 2.



79

TA

BLE

3S

ticki

ng E

ffici

ency

α

for

Viru

ses

from

Col

umn

Stu

dies

(6–

35 c

m)

Viru

sS

oil

pHN

aCl [

M]

f oc

�R

ef.

MS

2G

lass

bea

ds5.

00.

01—

0.00

15B

ales

et

al.

(199

1)a

5.0

0.08

—0.

0015

– 0

.002

8S

ilica

bea

ds5.

00.

052.

00 ×

10–7

0.01

8B

ales

et

al.

(199

3)7.

00.

052.

00 ×

10–7

0.00

22S

and

(Cap

e C

od)

5.7

0.1

<1.

0 ×

10–4

0.00

7K

inos

hita

et

al.

(199

3)7.

00.

1<

1.0

× 10

–40.

018.

20.

1<

1.0

× 10

–40b

Qua

rtz

3.5

0.01

—0.

12P

enro

d et

al.

(199

6)3.

50.

3—

0.16

5.0

0.01

—0.

009

5.0

0.1

—0.

095.

00.

3—

0.04

PR

D1

Gla

ss b

eads

5.5

0.01

—0.

0015

– 0

.003

3B

ales

et

al.(

1991

)a

San

d (B

orde

n)6.

50.

13.

0 ×

10–4

0.17

Kin

oshi

ta e

t al

. (1

993)

7.0

0.1

3.0

× 10

–40.

147.

60.

13.

0 ×

10–4

0.15

San

d (C

ambr

idge

)7.

00.

15.

0 ×

10–4

1.11

San

d (C

ape

Cod

)5.

70.

1<

1.0

× 10

–40.

62 –

≥ 0

.94

7.0

0.1

<1.

0 ×

10–4

0.63

– 0

.82

8.2

0.1

<1.

0 ×

10–4

0.58

Pol

io 1

Sili

ca b

eads

5.5

0.05

2.00

× 1

0–70.

014

Bal

es e

t al

. (1

993)

7.0

0.05

2.00

× 1

0–70.

0040

7.0

0.05

2.00

× 1

0–70.

0072

λQ

uart

z3.

90.

01—

1.25

Pen

rod

et a

l. (1

996)

5.0

0.01

—0.

045

5.0

0.1

—0.

535.

00.

3—

0.65

aC

orre

cted

for

the

fac

tor

As1/

3 , w

hich

was

not

app

lied

in t

he o

rigin

al p

ublic

atio

n.b

Set

to

0, b

ecau

se n

o re

mov

al w

as f

ound

.



80

TA

BLE

4S

ticki

ng E

ffici

enci

es

α fo

r V

iruse

s fr

om F

ield

Stu

dies

Viru

sS

oil

Dis

tanc

e [m

]pH

EC

[�

S/c

m]a

f oc

�R

ef.

MS

2S

and/

grav

el7.

57.

228

8—

0.00

4 –

0.18

2bD

eBor

de e

t al

. (1

999)

(Mis

soul

a)19

.47.

228

8—

0.00

4 –

0.20

2b

Dun

e sa

nd2.

47.

3 –

8.3

900

1.0

× 10

–30.

0014

Sch

ijven

et

al.

(199

9)30

7.3

– 8.

390

01.

0 ×

10–3

0.00

027

PR

D1

San

d0.

947.

431

13.

0 ×

10–4

0.00

28 –

0.0

030

Bal

es e

t al

. (1

997)

(Bor

den)

0.94

8.4

311

3.0

× 10

–40.

0008

5 –

0.00

16S

and

1.0

5.0

– 5.

760

– 1

00<

1.0

× 1

0–40.

009

– 0.

013

Pie

per

et a

l. (1

997)

(Cap

e C

od)

1.0

6.0

– 6.

735

0 –

450

1.0

× 10

–20.

0014

– 0

.002

6S

and

0.9

– 1.

05.

4 –

5.6

30 –

40

< 1

.0 ×

10–4

0.03

2 ±

0.01

6R

yan

et a

l. (1

999)

(Cap

e C

od)

0.9

– 1.

05.

8 –

6.0

250

– 33

01.

0 ×

10–2

0.01

6 ±

0.00

5S

and/

grav

el7.

57.

228

8—

0.01

4 –

0.63

2bD

eBor

de e

t al

. (1

999)

(Mis

soul

a)19

.47.

228

8—

0.00

5 –

0.38

5b

Dun

e sa

nd2.

47.

3 –

8.3

900

1.0

× 10

–30.

0024

Sch

ijven

et

al.

(199

9)30

7.3

– 8.

390

01.

0 ×

10–3

0.00

043

ϕX17

4S

and/

grav

el7.

57.

228

8—

0.00

6 –

0.31

1bD

eBor

de e

t al

. (1

999)

(Mis

soul

a)19

.47.

228

8—

0.00

7 –

0.31

9b

Pol

io 1

San

d/gr

avel

7.5

7.2

288

—0.

047

– 2.

108b

DeB

orde

et

al.

(199

9)(C

HA

T)

(Mis

soul

a)19

.47.

228

8—

0.01

9 –

0.86

6b

FR

NA

PH

’sD

une

sand

27.

3 –

8.3

900

—0.

0020

Sch

ijven

et

al.

(199

8)(0

.001

5 –

0.00

29)c

47.

3 –

8.3

900

—0.

0007

8(0

.000

040

– 0.

0040

)c

aC

orre

spon

ding

to

a ra

nge

in g

rain

siz

e fr

om 0

.001

25 –

0.0

12 m

.b

Ele

ctric

al c

ondu

ctiv

ity a

s a

mea

sure

of

ioni

c st

reng

th.

cN

inet

y-fiv

e pe

rcen

t co

nfid

ence

inte

rval

.



81

efficiency of MS2 is plotted vs. pH within the range of 4 to 9 and at different ionicstrengths, using the data from Table 3. The pI of MS2 is 3.5; it is thereforenegatively charged within this pH range. The different column studies appear toshow a coherent pattern. The value of α decreases with increasing pH, but reachesa minimum at a pH of about 5.5; then at higher pH, α remains constant. A similarpattern of electrophoretic mobility of MS2 vs. pH has been shown by Penrod et al.(1995), strongly indicating that surface charge of the virus particles as a functionof pH determines attachment. This pattern also strongly resembles that of the ratiokatt/kdet of group I viruses with pH (Figure 7). It also resembles that of the zetapotential of quartz as a function of pH (Loveland et al., 1996). The negative chargeof quartz decreases from pH 2.3 to 6, and remains minimal at pH values above 6.

MS2 did not adsorb or only poorly adsorbed in columns with sandy soils at pH5 to 8 (Bales et al., 1991, 1993; Kinoshita et al., 1993). Detachment was slow, butwas enhanced when pH increased. This effect was moderate, because few MS2particles had attached (Kinoshita et al., 1993). Bales et al. (1993) showed thatremoval of MS2 by attachment to silica beads in columns was much greater at pH5 than at pH 7. Also, poliovirus 1 attached much more at pH 5.5 than at pH 7. Bothviruses detached only slowly. At pH 7, the sticking efficiency of poliovirus 1 was2 to 3 times higher than that of MS2, but almost the same at pH 5 to 5.5. Penrodet al. (1996) compared the removal of bacteriophages MS2 and λ. Although the netsurface charge of λ was stronger negative than that of MS2, attachment of λ wasmore than that of MS2. It was suggested by Penrod et al. (1996) that a stericrepulsive force exists in the case of MS2. This steric repulsive force arises from thehydrophilic polypeptide loops, which extend a maximum of 1 nm off of the MS2

FIGURE 8. Sticking efficiency α vs. pH of MS2 at different ionic strengths. See also valuesin Table 3.



82

surface. A similar difference in attachment between MS2 and recombinant Norwalkvirus particles in quartz bed columns was found by Redman et al. (1997). MS2 hasa significantly lower pI but a less negative zeta potential at neutral pH and higherthan the Norwalk virus particles. Still, MS2 attached less than the Norwalk virusparticles at both pHs 5 and 7.

The sticking efficiency of PRD1 has also been shown to be pH dependent. Incolumns with silica beads at pH 5.5 (Bales et al., 1991), α for PRD1 was between0.006 and 0.013, and at pH 7, PRD1 did not adsorb at all (Table 3). Detachmentwas very slow compared with attachment, but increased strongly at pH 8. In thesame study, MS2 attached even less than PRD1. In a sandy aquifer under naturalgradient conditions, attached PRD1-phages released rapidly when pH was in-creased from 5.7 to pH 6 to 8 (Bales et al., 1995). Also, injection with high-pH-water in a sandy aquifer, which elevated pH from 7.4 to 8.4, resulted in significantremobilization of attached PRD1-phages (Bales et al., 1997). At pH 8.4, attach-ment rates decreased by a factor of 2 to 3. Concurrently, the detachment ratecoefficient increased by a factor 104 to 105 to a value higher than the attachmentrate coefficient. In a field study by Ryan et al. (1999), it was also shown that anincreased pH induced extra detachment of PRD1. This effect was less in a sewage-contaminated zone, where a pH of 8.5 was reached, than in an uncontaminatedzone, where a pH of 10 was reached. According to Ryan et al. (1999) an increaseto pH 8.5 may not exceed the isoelectric point of some ferric oxyhydroxides. It islikely that in the study by Bales et al. (1997) phosphate adsorbed to the ferricoxyhydroxides thereby augmenting the charge reversal caused by the pH increase.

Also, Loveland et al. (1996) clearly showed that at higher pH, attachment ofPRD1 is slower and detachment is faster. In this study, in equilibrium nonflowingcolumns, PRD1 was attached to quartz and ferric oxyhydroxyde-coated quartzsurfaces over a wide range of pH values. At a pH value of about 2.5 to 3.5 pH unitsabove the isoelectric point of the mineral surfaces, a so-called attachment edgeexisted. Viruses attached at a pH below this edge were irreversibly attached butwere reversibly attached at a pH above this edge. DLVO potential energy calcu-lations showed that the attachment edge occurred at the pH at which the potentialenergy of the primary minimum was near zero, implying that the position of theprimary minimum (attractive or repulsive) controlled the equilibrium distributionof the viruses. Virus attachment to a homogeneous mineral surface is eitherirreversible or reversible, depending on the pH of attachment, but not both (Lovelandet al., 1996). However, in a heterogeneous medium, minerals may exist to whichattachment is irreversible (e.g., ferric oxyhydroxide, and minerals to which attach-ment is reversible such as quartz). This is the case when pH lies between that ofthe isoelectric points of the two types of sites (Loveland et al., 1996).

From batch studies it has become clear that high pH generally favors free virusand low pH favors attached virus (Gerba, 1984). This has been confirmed incolumn and field studies for viruses like MS2 and PRD1 that have low pIs, but alsofor poliovirus 1 (pI of 6.6), and phage λ (pI of 5). At higher pH, electrostatic



83

repulsion increases, reflected in a higher maximum and a higher primary minimumin the DVLO profile. This results in a decreased attachment rate and an increaseddetachment rate. In most aquifers, surface characteristics of the soil are much moreheterogeneous than that of the quartz sand in the columns of Loveland et al. (1996),and also different viruses may be present that have different pIs. Therefore,depending on pH and thus on the charge of the virus and soil particles, adsorptionof some of these viruses may be irreversible, whereas that of others may bereversible.

E. Effect of Ionic Strength and Multivalent Cations

According to DLVO theory, a higher ionic strength compresses double layers,thereby increasing attachment rates. In batch studies, it was shown that viruses tendto adsorb strongly to various materials at high ionic strength (Lipson and Stotzky,1983; Grant et al., 1993). In a column study by Penrod et al. (1996), it was foundthat the sticking efficiencies of bacteriophages MS2 and λ increase with ionicstrength in the range of 0.01 to 0.1 M NaCl at pH 5 (Table 3). A similar increaseof α for MS2 is the case at pH 3.5 (see also Figure 8). A further increase of saltconcentration to 0.3 M NaCl has little impact on the sticking efficiency of phageλ, but causes a decrease in the value of α for MS2. The findings of Dowd et al.(1998) suggest that more negatively charged viruses attach more than less nega-tively charged viruses in the case of high concentrations of multivalent cations.Redman et al. (1999) studied adsorption of a filamentous bacteriophage, SJC3, inquartz columns at different ionic strengths and cation valence. SJC3 is a bacte-riophage that has been isolated from highly treated wastewater. Experiments werecarried out at pH 7, where surface charge of the bacteriophage and the quartz arenegative. SJC3 attached more to the quartz when the ionic strength was increasedby increasing the concentration of a particular salt from 1 to 100 mM. Oneexception was observed when using 1 mM NaCl. Contrary to the expectation,removal of the bacteriophage was higher in this case. The authors suggested thatat this low concentration of NaCl the filamentous bacteriophage stiffened. Due tothis stiffness, SJC3 may be retained in the column by straining.

If DLVO profiles are repulsive at all separation distances, DLVO theorypredicts increased detachment rates at lower ionic strength (Ryan and Elimelech,1996). In agreement with this prediction, Gerba and Lance (1978) found thatdetachment of poliovirus 1 was enhanced by deionized water that was applied tosimulate heavy rainfall. Redman et al. (1999) also showed that attached particlesof bacteriophage SJC3 easily detach when the salt concentration was changed from1 mM CaCl2 to 1 mM NaCl. However, if the DVLO profile is attractive near thesolid surface, DLVO theory will predict an increasing detachment rate at higherionic strength (Ryan and Elimelech, 1996). Penrod et al. (1996) found that withincreasing ionic strength there was a similar increase of both the attachment and



84

detachment rate coefficients. This suggests that in the case of the experiments ofPenrod et al. (1996), forces between the virus and the solid surface were attractiveat a very short separation distance.

Multivalent cations can link virus and adsorbents of like charge by forming saltbridges between them (Sobsey et al., 1980; Moore et al., 1982; Lipson and Stotzky,1983) or by charge reversal (Grant et al., 1993). Bales et al. (1991) showed in abatch experiment at pH 5 that attachment of MS2 to silica beads was at least 10times higher in the presence of Ca2+ than without Ca2+. These effects of multivalentions are of electrostatic nature. Redman et al. (1999) also showed that bacterioph-age SJC3 attached more to quartz when the ionic strength was increased by usinga multivalent (Ca2+or Mg2+) instead of a monovalent (Na+) cation. In a field studyby Bales et al. (1997) with the same soil type, pH, and velocity as in the columnstudy of Kinoshita et al. (1993), PRD1 attached much less than in the columnstudy. Pieper et al. (1997) suggested that a possible explanation for the differencemight have been the presence of calcium phosphate affecting surface charge in thecolumn study of Kinoshita et al. (1993) by cation bridging or charge reversal.However, as Farrah (1982) showed, calcium phosphate, which is an antichaotropicagent, may also promote hydrophobic interactions.

F. Effect of Organic Matter

The major occurrence of dissolved and/or suspended organic matter is in theform of humic substances (Shimizu et al., 1998). Commonly, humic substancesare, similar to viruses, negatively charged, and hence they compete with viruses forthe same binding sites (Gerba, 1984). Among dissolved and/or solid organic matterin the aquifer matrix, humic substances have the highest affinity to nonionichydrophobic organic compounds (Ouyang et al., 1996; Shimizu et al., 1998). Otherforms of dissolved organic matter consist of proteins, polypeptides, and aminoacids. From equilibrium batch experiments, it has become clear that dissolvedand/or suspended organic matter tends to compete with viruses for attachment siteson the soil surface and thereby reduce virus attachment (Gerba, 1984). Dissolvedor suspended organic matter can also be used to elute previously attached viruses(Sobsey et al., 1980). Thus, detachment of viruses may be strongly increased bydissolved or suspended organic matter.

Dizer et al. (1984) showed that in sand columns at pH 7, attachment rates ofpoliovirus 1, coxsackieviruses A9 and B1, echovirus 7, and rotavirus SA11 werethe lowest in secondary effluent when compared with groundwater, tertiary efflu-ent, and distilled water. Dizer et al. (1984) also found that virus attachment to sandin a batch test was reduced by dissolved surfactants. It was suggested that thesesurfactants disrupted hydrophobic bonds between the sand and the viruses. Suchsurfactants may also have been present in the secondary effluent, which decreasedvirus attachment in the columns.



85

The effect of dissolved organic matter may not always be apparent (e.g., if thenature of the soil is more important for adsorption). Gerba and Lance (1978) foundthat adsorption of poliovirus 1 and its movement through a loamy sand soil was notaffected by the high organic matter content of primary sewage effluent (70 mg/l)compared with that of secondary sewage effluent (10 mg/l). According to theauthors, the sewage effluent did not saturate adsorption sites, even after 9 days offlooding. Thus, dissolved organic matter has not influenced adsorption of poliovi-rus 1 because an excess of adsorption sites was available.

Under unsaturated flow conditions, the effect of dissolved organic matter iseven more complex. Powelson et al. (1991) found that natural humic material andsewage sludge organic matter decreased removal of MS2 in unsaturated columnswith loamy sand. On the contrary, in a field study, Powelson et al. (1993) foundthat the effluent type did not affect removal of MS2, whereas PRD1 was removedthree times more in secondary treated effluent compared with tertiary treatedeffluent, which has lower concentrations of dissolved organic matter. Powelsonand Gerba (1994) compared behavior of poliovirus 1 and bacteriophages MS2 andPRD1 in columns with a sandy soil under unsaturated conditions with secondaryand tertiary effluent, but found no significant influence of the effluent type onremoval of any of these viruses.

Similarly, the effect of solid and/or bonded organic carbon content (foc) of thesoil is found to be ambiguous. On one hand, a soil with bonded organic carbon hasfewer sites for virus adsorption. On the other hand, the bonded and/or solid organicmatter may provide hydrophobic adsorption sites in a soil where virus adsorptionis otherwise very low. So, the effect depends on the combination of soil type, virustype, and nature of organic matter. Gerba et al. (1981) carried out a statisticalanalysis on the batch adsorption data from Goyal and Gerba (1979). Adsorption ofgroup-I viruses (Table 2) was inversely related with pH and with foc. Variation inadsorption was explained for 83% by pH and for 12% by foc. In batch experiments,Moore et al. (1982) observed that adsorption of reovirus 3 to 30 different types ofsoils and minerals was found to decrease significantly with increasing foc. How-ever, when the soil with the highest organic content was omitted from the analysis,there was no significant correlation with foc. The authors concluded that foc has asignificant effect only at high values. The value of foc of the organic muck was3.4%. So, probably, at low foc, organic matter does not occupy all adsorption sitesfor viruses. In another batch study, Moore et al. (1981) found a strong negativecorrelation between foc and adsorption of poliovirus 2 to 34 different types of soilsand minerals. The authors described that soil organic matter is a weak adsorbentfor poliovirus because of its low pI, and therefore negative charge, and becauseorganic matter exhibits weak van der Waals attraction. They further reasoned thatsome of the soil organic matter, particularly lower-molecular-weight factions, issoluble and may pass into solution. The dissolved organic matter may be adsorbedonto other surfaces, and thereby compete for the same binding sites as viruses. Onthe other hand, in batch studies reported by Burge and Enkiri (1978), adsorption



86

of ϕX174 to four different kinds of soil was found to increase with increasing foc.Nevertheless, adsorption of ϕX174 to a fifth soil with a much higher foc was foundto be much lower. According to the authors, this was due to blocking of adsorptionsites by organic matter. They did not explain the positive correlation of adsorptionwith foc when considering only the four soils. However, foc of these four soils wasalso negatively correlated with pH. Because the effect of pH on virus adsorptionmay be expected to be stronger than that of foc (Gerba et al., 1981), pH probablyacted as a confounder. That is to say, the observed effect on adsorption was ratheran effect of increasing pH than of decreasing foc.

An example of cases, where bonded organic material may provide hydropho-bic adsorption sites for viruses, is clearly shown by Bales et al. (1993). They usedcolumns with negatively charged silica beads at pH 7, which were artificiallycoated with very small amounts of hydrophobic C18-chlorosilane. Attachment ofMS2 could be increased considerably by even very small amounts of this bondedhydrophobic organic carbon material (Table 4 and Figure 9). Bales et al. (1993)also showed that the binding sites in a column of silica beads with only 0.00215%bonded C18-chlorosilane could not be saturated with even 70 pore volumes of 105

MS2 particles per ml. The concentration of bonded MS2 particles was approxi-mately 107 pfu/g.

Yet, an opposite effect was observed in a field study by Pieper et al. (1997) ina sand and gravel aquifer with positively charged iron-oxide coating. They showedthat attachment of PRD1 was less in soil with higher foc. They injected 32P-labeledPRD1 into sewage-contaminated (foc of 1%) and uncontaminated zones (foc < 0.01%)

FIGURE 9. Sticking efficiency α of MS2 in 15-cm columns of silica beads (pH 7) with anincreasing fraction bonded organic matter, foc (C18-chlorosilane). Data from Bales et al.(1993). The highest value of α is actually > 0.016.



87

of the aquifer (see also data in Table 4). In the contaminated zone, only 42% ofPRD1 attached over the first meter, whereas in the uncontaminated zone 83%removal was found. Corresponding estimates of sticking efficiencies were foundto be about six times smaller in the contaminated zone than in the uncontaminatedzone. Injection of linear alkylbenzene sulfonates (LAS) remobilized 87% of at-tached PRD1 in the contaminated zone, but only 2.2% in the uncontaminated zone.The authors suggested that LAS caused charge reversal of ferric oxyhydroxidecoatings on the soil surface to which PRD1 was attached. They reasoned that theinjected amount of LAS was insufficient to cause this charge reversal in theuncontaminated zone. However, removal of LAS itself in both zones was similar(12 to 14%), which implies that this explanation is probably not correct. In anotherstudy at the same field site, Ryan et al. (1999) showed that anionic surfactantdodecylbenzenesulfonate (DBS) was also much more effective at mobilizing PRD1in the contaminated zone than in the uncontaminated zone. They argued thatabundant organic matter in the contaminated zone reduced the amount of DBSneeded to reverse the charge of ferric oxyhydroxides. Their results suggested thatPRD1 attached strongly in the uncontaminated zone.

Another explanation is that PRD1 in the uncontaminated zone primarily at-tached to sites with ferric oxyhydroxide coatings, whereas these sites were alreadytaken by organic matter in the contaminated zone. Therefore, attachment of PRD1in the contaminated zone may have been through mainly hydrophobic interactionswith bonded organic matter. This kind of attachment is weaker than attachment tosites of positively charged iron oxides (Moore et al., 1981). Thus, it may bereasoned that in the study by Pieper et al. (1997) and Ryan et al. (1999) injectedLAS and DBS, respectively, increases detachment of PRD1 in the contaminatedzone much more than in the uncontaminated zone by disrupting hydrophobic bondsbetween the soil and PRD1. This mechanism is consistent with the suggestion ofDizer et al. (1984) that surfactants can disrupt hydrophobic bonds (see also Gerba,1984). In the vicinity of a hydrophobic site, one may expect less electrostaticrepulsion, especially if that site consists of bonded polymeric organic matter. Suchpolymers may interact in two ways: by steric hindrance or by forming bridgesbetween surfaces. According to Ryan and Elimelech (1996), surfactant attachmentmay cause colloid release by some “non-DLVO” force like steric repulsion.

To summarize this section on the effect of organic matter, dissolved organicmatter may compete for the same binding sites as viruses, and because they areusually present in higher concentrations than viruses, they can decrease virusattachment. Dissolved organic matter, like surfactants, may also disrupt hydropho-bic bonds between soil and virus, resulting in an increased detachment rate. At thesame time, viruses and many organic materials contain hydrophobic groups ontheir surfaces. Therefore, once adsorbed, organic matter may provide hydrophobicbinding sites for viruses. Also, the presence of solid organic matter may result inan increased attachment rate through hydrophobic binding. The hydrophobic



88

adsorption effect will be most pronounced in soil with negatively charged grainsurfaces. The enhancing and attenuating effects of organic matter on adsorption ofviruses and their dependence on soil and virus properties make their quantificationvery difficult, especially under field conditions. Effects of organic matter are,probably, responsible for considerable uncertainty in predicting virus removal.

V. VIRUS INACTIVATION

A. Introduction

In Equations 3 and 4 virus inactivation was described by the inactivation ratecoefficients µl and µs of free and attached viruses. Modeling of the inactivation offree viruses is described in Section V.B. Section V.C discusses the modeling of µs

in a batch suspension with soil. In Section V.D modeling of virus inactivationduring subsurface transport is discussed.

B. Virus Inactivation in Water

Virus inactivation is usually assumed to follow first-order kinetics, as inEquations 3 and 4. However, nonlinear inactivation curves, resulting from externaland internal factors, have also been found. External factors that influence virusinactivation are the environmental factors that will be discussed in Section IV andaffect the population of viruses as a whole. Hurst et al. (1992) analyzed the effectsof different environmental factors on virus inactivation by means of differentregression equation formats and evaluated these formats. Hurst et al. (1992)considered environmental factors such as temperature, conductivity, and turbidityof the water, and ability to support bacterial growth. According to Hurst et al.(1992), virus inactivation is not a first-order but a time-dependent process.

Internal factors that influence virus inactivation may be the presence of virusaggregates or the existence of significant variations in sensitivities among the viruspopulation to the factors that cause inactivation (Yates et al., 1987). Grant (1994)presented a mathematical model for first-order inactivation of total infectious unitsof viruses, including Brownian coagulation of virions. This model predicted thatvirion–virion coagulation is negligible in most aquatic environments because virusconcentrations are too low or because inactivation occurs too quickly. The pre-dicted coagulation in highly concentrated laboratory suspensions was too highalso, because laboratory studies had indicated that at neutral or alkaline pHcoagulation of poliovirus was minimal. Grant (1994) reasoned that to the extentthat viral aggregates actually exist in aquatic environments, they are probablyformed intracellularly and not by coagulation outside the host cell. Furthermore, inaquatic environments, virus aggregates may be dispersed by changes in solution



89

chemistry. Grant (1995) also developed a kinetic model for the inactivation ofviruses that exhibit a range of initial aggregate sizes. Aggregates of viruses areassumed to be more resistant to inactivation because all virus particles within anaggregate must be inactivated before the aggregate as a whole is consideredinactive. Also, undamaged components of inactive virus particles within an aggre-gate may recombine by a process called multiplicity reactivation (MR). Grant’skinetic model successfully fitted experimental data from a study on inactivation ofvaccinia virus. Usually data on inactivation of individual virus particles, as well asthe level of initial aggregation and the number of active components per virusparticle needed for MR are not available, but in this study on the inactivation ofvaccinia virus they were. The model correctly predicted the relatively slow inac-tivation that was observed for the aggregated suspension. None of the models ofGrant (1994, 1995) address higher-order inactivation, due to variation in individualviruses within a population of viruses, association with other types of viruses, orassociation with other nonviral organic or inorganic material. These associations,however, are known to be important (see Section VI.C).

Grant et al. (1993) showed a biphasic decline of the logarithmic concentrationwith time of phage λ, indicating that a population of phage λ consists of twosubpopulations with different sensitivity toward inactivation. Rossi (1994) arguedthat such a biphasic course of inactivation was possibly caused by interaction ofvirus particles with the walls of the synthetic vials that were used. He showed thatthis was the case for phage T7, but in a glass vial its inactivation was first order.He also mentioned the existence of very fine colloidal clay particles, to which theviruses were attached, as a possibility that may have caused nonlinear inactivation.

C. Virus Inactivation in Water with Soil

According to Grant et al. (1993), batch experiments involving viruses and soilare fundamentally nonequilibrium systems, because viruses will also disappearfrom the liquid phase by inactivation. Inactivation of viruses that are attached tosoil may proceed at a different rate than that of free viruses. Grant et al. (1993)developed a numerical model that enables estimation of katt, kdet, µl, and µs forviruses in a batch suspension with soil, and demonstrated this for phage λ. No otherstudy has been reported where values of each of katt, kdet, µl, and µs were estimatedby kinetic analysis of batch experiments. In a batch system, viruses are subject toadsorption and inactivation, simultaneously. Depending on the relative strength ofthese processes, Grant et al. (1993) distinguished the following four possiblebehaviors (Figure 10):

1. Quasi-equilibrium adsorption (QEA) is the case when virus inactivation is notinfluenced by the presence of a solid; thus, the inactivation rate coefficient ofdetached viruses equals that of attached viruses (µl = µs).



90

FIGURE 10. Semi-log plot showing four possible kinetic behaviors in a batch suspensionof viruses with soil compared with pure inactivation of viruses in a suspension without soil(Grant et al., 1993). C/C0 is reduced concentration of viruses in aqueous phase; τ isnondimensional time. See Section V.C for an explanation of behaviors. QEA: µs = µl; QEARI:µs < µl; QEASS: µs > µl; IA: kdet = 0.

2. Quasi-equilibrium adsorption and reduced inactivation (QEARI) is observedwhen virus particles that are attached to solid materials inactivate slower thanthose in the aqueous phase (µs < µl).

3. Quasi-equilibrium adsorption and surface sink (QEASS) is the case when thesolid surface acts like a sink for virus particles if virus inactivation is eitheraccelerated in the presence of the surface (µs > µl), or a portion of reversiblyattached viruses becomes irreversibly attached with time. These two possibili-ties are indistinguishable.

4. Irreversible attachment (IA) is observed when viruses are attached irreversiblydirectly from solution (kdet = 0). The curves for IA and pure inactivation havethe same intercepts, but the slope of the IA curve is steeper (Figure 10).

There is another way of examining inactivation rate coefficients from batchexperiments, as demonstrated by Rossi (1994) and Formentin et al. (1997). Virusesthat are attached to clay particles are still capable of infecting their host, and thuscan still be enumerated. In such batch experiments with clay particles, it is thuspossible to count the total number of viruses (i.e., the number of attached and freeviruses together from samples taken at different times). By subtracting the countednumber of free virus particles from that of the total number, the number of attachedinfectious virus particles is obtained. In this way, inactivation of free and attachedviruses can be followed and compared directly.



91

Commonly in batch studies, only the free virus concentration is measured withtime. Several of such studies exist that have compared inactivation of viruses in abatch suspension with and without soil added (Sobsey et al., 1980, 1986; Nasseret al., 1993; Gantzer et al., 1994; Blanc and Nasser, 1996; Sakoda et al., 1997).From the data of these studies it is still possible to calculate a value of µs. In thesestudies, the percentage of adsorption and the decay rate of the free virus concen-tration in the suspension with soil, µeff, were measured. The value of this decay ratelies between that of µl and µs. In these studies, an apparent equilibrium was reachedwithin a short period of time, whereas inactivation was measured over a muchlonger period of time. Therefore, it is possible to calculate µs from the ratio katt/kdet,µl and µeff. Applying Equation 10, but eliminating the terms for dispersion andadvection, and substituting Equations 9, 2, and 16 gives:

RdC

dtR Cl s= µ +( )µ( )– –1 (25)

This equation has the following solution:

C CR

Rtl s=

+( )

0

1exp –

–µ µ(26)

Now, µs can be calculated as follows:

µ = µ( ) ( ) =

s eff lR R

t

C

C– – – lnµ µ1

1

0

where eff (27)

Values of µs, calculated using Equation 27 from the batch studies of Sobsey et al.(1980, 1986) and Blanc and Nasser (1996), are presented and evaluated in SectionVI.

D. Virus Inactivation under Transport Conditions

Several virus transport models have been developed that incorporate constantfirst-order inactivation of free (µl) and attached (µs) viruses (Chrysikopoulos andSim, 1996; Sim and Chrysikopoulos 1995, 1996, 1998; Toride et al., 1995). Also,models exist that do not distinguish between inactivation of free and attachedviruses (Park et al., 1994, 1995; Tim and Mostaghimi, 1991; Yates et al., 1986;Yates and Ouyang, 1992; Yates and Yates, 1987, 1988). Based on the existence of



92

two or more subpopulations of viruses in a suspension with different inactivationrate coefficients, Sim and Chrysikopoulos (1996) also developed a transport model,incorporating kinetic reversible adsorption and different time-dependent inactiva-tion rate coefficients for suspended and attached viruses. The multiphasic sequen-tial inactivation was approximated by a pseudo-first-order expression with a time-dependent inactivation rate coefficient. This way, inactivation of viruses in somebatch studies was simulated better. Model simulations showed that this approxima-tion of virus inactivation implies extended survival of viruses, and consequentlymore distant migration. This more sophisticated way of modeling virus inactiva-tion during subsurface transport under saturated conditions may also be usefulwhen considering heterogeneous populations of different viruses that exhibitmultiphasic inactivation.

The contribution of virus inactivation to the removal of virus under unsaturatedconditions is much greater. Furthermore, daily fluctuations in temperature in theupper unsaturated soil layers are considerable. Therefore, Yates and Ouyang(1992) developed a model (VIRTUS) that simultaneously describes the transportof water, heat, and viruses through the unsaturated zone of the soil. VIRTUS wasonly tested with data from one unsaturated column experiment (Powelson et al.,1990), where model predictions were close to measured breakthrough concentra-tions. The temperature-dependent inactivation rate capabilities of the model werenot tested, simply because experiments were carried out at constant temperature.

In experiments with saturated columns, inactivation of viruses is usuallyinsignificant, and therefore, neglected within the time scale of the experiment(Bales et al., 1989, 1991, 1993; Dowd et al., 1998; Jin et al., 1997; Penrod et al.,1996; Redman et al., 1997). At field scale, virus inactivation may be significantbecause of much longer timescales. To date, the field study by Schijven et al.(1999) is the only field study where both µl and µs were estimated (Table 8). Theinactivation rate coefficient µl for MS2 and PRD1 was estimated from inactivationrates in suspensions with recharge water from the field. The inactivation ratecoefficient µs for each of these phages could be estimated from the slope of the tailsof the breakthrough curves (Figure 3).

VI. FACTORS AFFECTING VIRUS INACTIVATION IN THESUBSURFACE

A. Introduction

Viruses lose their ability to infect host cells with time by inactivation. Virusesare inactivated because of disruption of coat proteins and degradation of nucleicacids (Gerba, 1984). The factors that influence inactivation of viruses have alreadybeen reviewed by Yates et al. (1987). They mention three reports on inactivationrates for viruses in groundwater (Keswick et al., 1982; Bitton et al., 1983; Yates



93

et al., 1985). Since then a number of new studies have been carried out. Table 5shows an inventory of the values of µl for viruses in groundwater and sewage waterfrom several studies. Table 6 gives an inventory of µs-values that were calculatedusing Equation 27 and data from the studies of Sobsey et al. (1980, 1986) andBlanc and Nasser (1996). In some cases, the value of Rµeff was less than that of µl.This would result in an estimate for (s of less than zero, which is unrealistic. Theseestimates were, therefore, omitted. Possibly, the value of the ratio katt/kdet washigher than measured. This implies that equilibrium was not reached in these cases,because adsorption was too slow. As already pointed out in Secton III.C, this leadsto an underestimation of the ratio katt/kdet, and thus of R.

Inactivation is usually regarded as a first-order process (see Section V). Themost important factors that influence virus inactivation rates are temperature,adsorption to particulate matter and soil, unsaturated conditions, and microbialactivity. These factors are discussed in the following sections.

Effects of other factors are found to be of insignificant importance, althoughsome exceptions are reported. Yates et al. (1985) found that pH, ammonia, calciumhardness, magnesium hardness, total hardness, nitrate, total dissolved solids, andturbidity did not significantly affect inactivation of MS2, poliovirus 1, and echo-virus 1. Inactivation of MS2, however, increased significantly with calcium hard-ness. Mathess et al. (1988) reported that the inactivation rates of coxsackievirus A9and B1, echovirus 7, and poliovirus 1 were larger in deionized than untreatedgroundwater. The type of water (groundwater, secondary, and primary effluent)was not found to significantly affect inactivation rates of poliovirus 1, echovirus 1,HAV, MS2, and PRD1 (Sobsey et al., 1986; Blanc and Nasser, 1996). However,Nasser et al. (1993) found a greater value of µl of poliovirus1, HAV, and F-specificRNA bacteriophages (FRNAPH’s) in primary effluent than in groundwater.

B. Effect of Particulate Matter and Soil on Virus Inactivation

Viruses in the environment are often associated with particulate matter or othersurfaces and this has a major effect on their inactivation and transport in theenvironment (Gerba, 1984). Adsorption to organic matter or clays, like kaoliniteand montmorillonite, or aggregation in the environment has been demonstrated forbacteriophages T2 (Gerba and Schaiberger, 1975), T7 (Bitton and Mitchell, 1974),and f2 (Armon and Cabelli, 1988) and for poliovirus 1 (Landry et al., 1983; Metcalfet al., 1984) and poliovirus 2 (Moore et al., 1981). Natural colloids, which areusually defined by their size of 1 to 1000 nm, are present in relatively largeconcentrations (ranging from 108 to 1017 particles per liter) in waters of diversegeological environments (Swanton, 1995). Gerba et al. (1978) found that in treatedsewage effluent, the largest quantity of solid-associated coliphages are attached toparticles greater than 8.0 (m and less than 0.65 µm. However, Hejkal et al. (1981)indicated that in treated wastewater the majority of enteric viruses are free or



94

TA

BLE

5In

activ

atio

n R

ate

Coe

ffici

ent

�l [

day

–1]

of V

iruse

s in

Gro

undw

ater

and

Was

tew

ater

°CW

ater

MS

2F

RN

AP

Hs

PR

D1

Pol

io 1

Ech

o 1

HA

VR

ef.

4G

RW

0.06

3Y

ates

et

al.

(198

5)b

7G

RW

0.00

580.

01Y

ahya

et

al.

(199

3)7

GR

W0.

100.

1010

GR

W0.

110.

025

0.11

0.19

Bla

nc a

nd N

asse

r (1

996)

10G

RW

a0.

010

Mat

thes

s et

al.

(198

8)10

GR

W-d

eion

ized

0.03

210

GR

W0.

013

10G

RW

00

0.10

Nas

ser

et a

l. (1

993)

10P

Ea

00.

046

0.17

10P

E0.

031

0.00

770.

1210

SE

0.07

70.

054

0.03

Bla

nc a

nd N

asse

r (1

996)

10T

E0.

091

0.05

10.

110.

1712

GR

W0.

160.

180.

24Y

ates

et

al.

(198

5)b

13G

RW

0.22

0.20

0.25

Yat

es e

t al

. (1

985)

b

16G

RW

(5.

4 m

g/l O

2)0.

21Ja

nson

s et

al.

(198

9b)

16G

RW

(0.

2 m

g/l O

2)0.

069

17G

RW

0.18

0.31

0.28

Yat

es e

t al

. (1

985)

b

20G

RW

0.00

770.

038

0N

asse

r et

al.

(199

3)20

PE

a0.

038

0.07

70.

1520

PE

0.08

40.

140.

1520

SE

a0.

054

Sob

sey

et a

l. (1

980)

20S

E0.

1422

GR

W (

0.06

mg/

l O2)

0.16

Jans

ons

et a

l. (1

989)

22G

RW

0.10

Bitt

on e

t al

. (1

983)

23G

RW

0.36

0.03

50.

170.

18B

lanc

and

Nas

ser

(199

6)23

GR

W1.

30.

12Y

ahya

et

al.

(199

3)23

GR

W0.

580.

3023

GR

W0.

731.

20.

92Y

ates

et

al.

(198

5)b

23S

E0.

380.

180.

230.

025

Bla

nc a

nd N

asse

r (1

996)

23T

E0.

280.

069

0.15

25G

RW

a0.

082

Sob

sey

et a

l. (1

986)

25G

RW

0.33

25P

Ea

0.10

0.10

> 0

.055

25P

E0.

330.

330.

055



95

25S

Eb

0.13

0.08

20.

055

25S

E0.

130.

130.

073

30G

RW

0.03

10.

120.

054

Nas

ser

et a

l. (1

993)

30P

Ea

0.03

80.

120.

2030

PE

0.01

50.

210.

18

° CW

ater

Reo

3C

A9

CB

1C

B5

Ech

o 6

Ech

o 7

Ech

o 11

Ech

o 24

Ref

.

10G

RW

a0.

019

0.01

20.

032

10G

RW

0.02

70.

019

0.01

9M

atth

ess

et a

l. (1

988)

10G

RW

-dei

oniz

ed0.

031

0.04

00.

038

16G

RW

(2.

3 m

g/l O

2)0.

23Ja

nson

s et

al.

(198

9b)

16G

RW

(1.

6 m

g/l O

2)0.

1219

GR

W (

1.2

mg/

l O2)

0.12

20S

Ea

0.12

Sob

sey

et a

l. (1

980)

20S

E0.

5122

GR

W (

0.2

mg/

l O2)

0.25

Jans

ons

et a

l. (1

989b

)

Not

e:

GR

W,

grou

ndw

ater

; P

E,

prim

ary

efflu

ent;

SE

, se

cond

ary

efflu

ent;

TE

, te

rtia

ry e

fflue

nt.

aS

teril

ized

bA

vera

ge v

alue

s.



96

TA

BLE

6In

activ

atio

n R

ate

Coe

ffici

ent

µ s o

f V

iruse

s in

Wat

er w

ith S

oil

Wat

er° C

k att/k

det

µ sµ s

k att/k

det

µ sµ s

k att/ k

det

µ sµ s

Sob

sey

et a

l. (1

980)

Pol

io 1

Reo

3

nst

stns

tst

Lake

land

san

dS

E20

3.2

0.08

60.

022

1.2

—0.

037

Pon

zer

muc

kS

E20

0.5

1.2

0.65

120.

250.

024

Ben

toni

teS

E20

2.3

0.12

0.05

6K

aolin

iteS

E20

990.

093

0.04

8

Sob

sey

et a

l. (1

986)

Pol

io 1

Ech

o 1

HA

V

nst

stns

tst

nst

st

FM

san

dG

RW

250.

470.

380.

18S

E25

5.7

0.23

0.06

30.

042

0.95

—0.

39P

E25

990.

130.

160.

064

—0.

293.

0C

orol

la s

and

GR

W25

0.06

41.

00.

51S

E25

0.52

0.71

0.02

10.

031

0.13

0.08

20.

064

PE

250.

720.

068

0.39

0.04

2—

0.39

0.12

Pon

zer

muc

kG

RW

250.

112.

21.

60.

087

2.1

1.6

0.69

—S

E25

0.07

50.

075

0.30

PE

250.

16—

0.96

0.12

—0.

200.

220.

21B

ento

nite

GR

W25

0.39

SE

250.

087

——

0.15

0.38

—0.

56P

E25

0.32

—0.

151.

5—

0.15

0.72

Kao

linite

GR

W25

99S

E25

990.

330.

065

990.

130.

060

990.

066

0.06

6P

E25

990.

220.

094

0.08

70.

33—

0.56

Cec

il sa

ndy

clay

GR

W25

990.

074

0.08

399

0.17

0.13

99S

E25

9999

99P

E25

990.

092

0.06

499

0.16

0.22

99



97

Bla

nc a

nd N

asse

r, 1

996

Pol

io 1

MS

2P

RD

1

nst

stns

tst

nst

st

Loam

y sa

ndG

RW

1019

90.

030.

180.

174.

80.

0723

0.16

0.46

0.15

SE

1014

20.

030.

180.

2715

0.09

230.

171.

340.

12S

and

GR

W10

3.5

0.01

911

0.30

370.

030

230.

022

0.44

0.08

1S

E10

0.38

0.01

33.

30.

1619

90.

088

230.

014

0.64

0.13

Not

e:G

RW

, gr

ound

wat

er;

SE

, se

cond

ary

efflu

ent;

PE

, pr

imar

y ef

fluen

t;. n

st,

not

ster

ilize

d; s

t, s

teril

ized

.



98

attached to particles smaller than 0.3 µm. Metcalf et al. (1984) also showed thatenteroviruses and rotaviruses in estuarine water adsorb preferentially to particlessmaller than 0.3 µm in diameter. Particles less than 0.3 µm include clays, cellfragments, waste products, and other miscellaneous debris (Levine et al., 1985).Furthermore, Payment et al. (1988) showed that in river water 77% of indigenousenteric viruses and 66% of coliphages were probably free or associated withparticles with a diameter of less than 0.25 µm. Thus, a substantial fraction ofviruses is attached to wastewater effluent solids and other colloidal particles witha size of less than 0.3 µm.

Virus association with solids in natural waters has been observed to generallyreduce virus inactivation, but some exceptions have been reported too (Gerba,1984). Reduced inactivation due to adsorption to suspended matter or marinesediment in seawater or estuarine water has been reported for poliovirus 1, echo-virus 1, coxsackieviruses A9 and B3, rotavirus SA11, HAV, FRNAPHs, andphages of Bacterioides fragilis (Smith et al., 1978; Liew and Gerba, 1980; Metcalfet al., 1984; Rao et al., 1984; Chung and Sobsey, 1992; Gantzer et al., 1994).Mathess et al. (1988) showed that addition of sand to groundwater at 10°C reducedthe inactivation rate of coxsackievirus B1, but the inactivation rates of coxsackievirusA9 and echovirus 7 were not affected by the addition of sand. Grant et al. (1993)showed a reduced inactivation of phage λ in a batch suspension with Ottawa sandat pH 10, but not at pH 5 or 7. Sakoda et al. (1997) also reported significantlyreduced inactivation of bacteriophages MS2 and Qβ in PBS (pH 7.2) by additionof cellulose, DEAE-cellulose, cellulose phosphate, kaolin, carbon, suspended sol-ids, and sediment from a river.

In some studies increased inactivation in the presence of soil was reported.Sobsey et al. (1980) found this for poliovirus 1 in Ponzer muck. Blanc and Nasser(1996) reported this for MS2 and PRD1 in sand and loamy sand. In the study ofFormentin et al. (1997), it was observed that bacteriophage f1, a filamentous phagewith a length of 100 nm, attached massively to attapulgite and was subsequentlyinactivated more rapidly. In this study, it was found that concentrations of free andattached MS2 particles increased after the addition of attapulgite. Possibly, exist-ing aggregates of MS2 disintegrated when attachment to attapulgite took place.

Reduced inactivation was found to be more pronounced in case of strongerattachment, especially to clays (Hurst et al., 1980). Babich and Stotzky (1980)found that in the presence of clay minerals, the inactivation rate of phage ϕ11M15in natural lake water was greatly reduced, with the sequence of protection beingattapulgite > vermiculite > montmorillonite > kaolinite. Whether reduced inactiva-tion in the presence of clay will be observed may also depend on its concentration.Gerba and Schaiberger (1975) did not find any increase in the survival duration ofphage T2 in the presence of low kaolinite concentrations in natural seawater. Also,Gantzer et al. (1994) showed that the inactivation rate of poliovirus 1 in seawaterwas not significantly reduced in the presence of low concentrations of montmoril-lonite (3 and 15 mg/l) compared with that without this clay. These low concentra-



99

tions of montmorillonite are representative of the density of suspended matter ina natural seawater. However, with 500 mg/l of clay, virus inactivation was signifi-cantly less.

In Figure 11, the ratio µeff/µs is plotted as a function of the ratio katt/kdet usingthe data from the studies that are summarized in Table 6. Within this context, avalue of µeff/µl of less than one defines reduced inactivation (see Equation 27). Itappears that inactivation of poliovirus 1, echovirus 1, and HAV is usually reducedbut in some cases enhanced, independent of katt/kdet. Inactivation of reovirus 3 isreduced in all cases, but that of MS2 is enhanced in all cases, and that of PRD1 isenhanced in most cases. In Figure 12, µs is plotted as a function of the ratio katt/kdet

using the data from Table 6. It also seems that µs is independent of katt/kdet. At valuesof katt/kdet < 1, the values of µs for poliovirus 1, echovirus 1, and MS2 show morevariation and can be relatively high. For values of katt/kdet > 1, the value of µs isgenerally less than about 0.25 day–1 for all viruses that are shown here. The datafrom Figures 11 and 12 suggest that reduced or enhanced inactivation is very muchvirus-type specific, and almost independent of katt/kdet.

Gerba (1984) mentioned that reduced inactivation may be the result of protec-tion from proteolytic enzymes or other virus-inactivating substances, increasedstability of the virus capsid when attached, prevention from aggregate formation,and blocking from ultraviolet radiation. The latter is of importance in surfacewaters. Sakoda et al. (1997) suggested that attachment of viruses onto a solidsurface prevents them from swelling, and subsequent activity loss.

In the studies analyzed above, the batch suspension was not agitated continu-ously. Therefore, a protection against forces along the air–water interface probably

FIGURE 11. Reduction factor for inactivation rate µeff/µl vs. katt/kdet for viruses in a batchsystem with soil.



100

FIGURE 12. Inactivation rate coefficient µs vs. katt/kdet for viruses in a batch system withsoil.

does not play a role. This may, however, play a significant role, depending on thedegree of agitation (see Section VI.E). In the cases where virus inactivation in thepresence of soil was enhanced, presumably this may have been caused by theattachment and detachment processes. Viruses may be degraded because of attach-ment to sites with metal oxides (Gerba, 1984).

Two field studies exist where virus inactivation was investigated during sub-surface transport under saturated conditions. In a field study by Schijven et al.(1999), the value of µl for free MS2 particles (0.03 day–1) was found to be threetimes less than the value of (s for attached MS2 particles (0.09 day–1), wherebyattachment was found to be very low. This observation is consistent with theanalysis from batch experiments with MS2 (Figure 11) that generally showedenhanced inactivation of MS2 in the presence of soil. In the case of PRD1, theestimate of (l was not clear, but was likely less than that of µs (0.07 day–1). On theother hand, DeBorde et al. (1998) found indications for reduced inactivation ofMS2 and ϕX174 during subsurface transport under field conditions. Inactivationwas followed for about one half year (6 to 12°C). In enclosed tubes, µl for MS2 andϕX174 were about 0.06 and 0.03 day,–1 respectively. The tails of the breakthroughcurves declined slower in time (i.e., 0.016 day–1 for MS2 and negligible forϕX174). There is no clear explanation for the difference between the two fieldstudies, where both enhanced and decreased inactivation of attached viruses undertransport conditions were observed.

C. Effect of Temperature

Temperature is the most important factor that influences virus inactivation(Hurst et al., 1980; Yates et al., 1985, 1987). Inactivation rates increase with



101

temperature (Hurst et al., 1980; Yates et al., 1985; Jansons et al., 1989b; Nasser etal., 1993; Yahya et al., 1993; Blanc and Nasser, 1996). Figure 13 shows how µl forMS2, FRNAPHs, PRD1, poliovirus 1, echovirus 1, and HAV changes with tem-perature in nonsterilized groundwater. This figure was constructed using the datafrom Table 5. Regression analysis suggests an exponential dependence of µl ontemperature in the form:

ln µ = +l aT b (28)

Values of coefficients a and b for six different viruses are given in Table 7. Here,it should be noted that this analysis was based on only a few data for FRNAPHsand PRD1; therefore, the values for these viruses are only indicative. Clearly, thetemperature sensitivity of µl depends on the type of virus. Yates et al. (1985) found

FIGURE 13. Inactivation rate coefficient µl vs. temperature for viruses in nonsterilizedgroundwater from different studies. See values in Table 5.

TABLE 7Regression Analysis of ln( �l) for Viruses with Groundwater Temperature(T,°C)

Equation: ln( �l) = aT + b Polio 1 Echo 1 HAV MS2 FRNAPHs PRD1

Number of observations 10 5 4 8 2 3Slope a 0.033 0.12 –0.024 0.12 0.14 0.09Intercept b –2.4 –3.0 –1.4 –3.5 –7.6 –4.0Correlation coefficient 7% 71% 8% 85% 100% 71%



102

no significant differences between inactivation rates of poliovirus 1, MS2, andechovirus 1 at different temperatures in groundwater samples from different sites.However, Sobsey et al. (1986) found lower values of µl for poliovirus 1 andechovirus 1 at 25 and 30°C. Therefore, the data in Table 7 indicate that inactivationof poliovirus 1 is much less sensitive to temperature than MS2 or echovirus 1 (seevalues of slope a). Table 7 further shows that HAV may be regarded as insensitiveto changes in temperature. FRNAPHs appear to be very stable as compared withother viruses, but are very sensitive to changes in temperature, as is MS2. Inseawater, inactivation rate coefficients for FRNAPHs and HAV were 0.08 and 0.07day–1 at 5°C, respectively, and 1.0 and 0.3 day–1 at 25°C, respectively (Chung andSobsey, 1992). So, possibly, FRNAPHs are not as stable as the data of Nasser etal. (1993) suggest.

The values of µs that were derived from Blanc and Nasser (1996) allowevaluation of µs variation with temperature. It appears that µs values for MS2,PRD1, and poliovirus 1 that are attached to loamy sand significantly increase withtemperature. The same is the case with µs values for MS2, PRD1 that are attachedto sand, but not for poliovirus 1. Probably, µs values change similarly withtemperature as µl values.

D. Effect of Microbial Activity

Inactivation of viruses may be enhanced because of microbial activity (Yateset al., 1987). Hurst et al. (1980) found that the concentration of sewage effluent indistilled water under sterile aerobic and anaerobic, as well as under nonsterileanaerobic incubation conditions did not significantly affect virus inactivation.However, under nonsterile aerobic conditions, inactivation of poliovirus 1 pro-ceeded faster. This indicates a deleterious effect of aerobic microorganisms on thesurvival of poliovirus 1. Also, inactivation of poliovirus 1 and reovirus 3 in settledsewage (Sobsey et al., 1980) and of poliovirus 1, echovirus 1, and HAV ingroundwater, secondary, and primary effluent (Sobsey et al., 1986) was slowerunder sterile conditions than under nonsterile conditions in most cases (Table 5).The same can be said for inactivation of attached viruses (Table 6). Jansons et al.(1989b) measured inactivation of poliovirus 1 in dialysis bags in different bore-holes at a field site. The dissolved oxygen concentration in groundwater wasdifferent between boreholes. They found that the inactivation rate of poliovirus 1was three times higher at a mean dissolved oxygen concentration of 5.4 mg/lcompared with 0.2 mg/l. Also, Pseudomonas maltophilia was found in largenumbers only in the borehole with the higher dissolved oxygen concentration.Inactivation of poliovirus 1 may have been directly affected by a higher dissolvedoxygen concentration, or by microbiological activity due to the higher dissolvedoxygen concentration.



103

Seemingly, inactivation rates of viruses are not always affected by the activityof microorganisms. Babich and Stotzky (1980) did not find a significant differencein inactivation rate of bacteriophage ϕ11M15 between natural, autoclaved, orfiltered lake waters. Also, Mathess et al. (1988) found no significant differencebetween the inactivation rates of coxsackievirus A9 and B1, echovirus 7, andpoliovirus 1 in sterile and nonsterile groundwater.

E. Effect of Unsaturated Conditions

In the calculation of attachment to soil in saturated columns, inactivation ofviruses is usually neglected within the time scale of the experiment. However,under unsaturated conditions, the contribution of the effect of inactivation on virustransport is much more important. Powelson et al. (1990) showed that MS2 wasneither adsorbed nor inactivated in a saturated 1-m column with a loamy fine sand(pH 8), because after less than two pore volumes the effluent concentration reachedthat of the influent concentration. However, in a column under unsaturated condi-tions, the effluent concentration was significantly reduced. By analysis of soilsamples from the unsaturated column, it appeared that MS2 adsorbed only poorly,but was removed by enhanced inactivation. Powelson and Gerba (1994) showedthat in columns under unsaturated conditions, removal of the MS2, PRD1, andpoliovirus 1 was more than three times higher as under saturated conditions.Poletika et al. (1995) showed in a lysimeter with undisturbed unsaturated soil thatMS2 was attached stronger and inactivated faster. These results suggested that theairwater interface may be retaining and/or inactivating viruses during transportthrough unsaturated soil.

Hurst et al. (1980) showed that the inactivation rate of poliovirus 1 increasedas the soil moisture content of a sandy soil increased from 5 to 15% and thendecreased when the soil moisture content increased from 15 to 25%. Apparently,inactivation was at its maximum near the soil moisture saturation point. Possibleexplanations for this observation would include soil moisture-level-dependentdifferences in the extent of virus attachment to the soil and the mechanisms ofattachment. Rossi (1994) showed that, in a batch test, inactivation of bacteriophageT7 increased due to the action of strong agitation. During this strong agitation,virus particles have a higher probability to come into contact with the air–waterinterface. Addition of organic matter (tryptone, humic acid) saturated the air–waterinterface with organic matter. This lowered the chance of a virus particle enteringthe air–water interface and resulted in a reduced inactivation rate. Addition ofattapulgite clay resulted in very fast attachment of the virus particles to the clayparticles, lowering the chance of entering the air–water interface even more and aneven stronger reduction of the inactivation rate was observed. Therefore, it wasbelieved that contact with the air–water interface increases inactivation. Also, byaddition of attapulgite, H40/1 and H6/1 were very effectively protected againststrong agitation (Formentin et al., 1997). Thompson et al. (1998) showed that



104

bacteriophage MS2 was protected against the air–water–solid contact line whenattached to soil particles. In this case, the solid was the hydrophobic polypropylenewall of the tube containing the suspension. At the air–water–polypropylene contactline surface tensions are much higher than at an air–water–glass contact line. Asimilar effect of a synthetic tube wall on inactivation was also observed by Rossi(1994) for phage T7 (see Section V.A). These findings suggest that physical forcesassociated with the air–water–solid contact line may be responsible for enhancedinactivation.

VII. ADVECTION AND DISPERSION OF VIRUSES

It has been shown in both laboratory and field studies that physical heteroge-neities of the soil increases the dissimilarities between the transport behavior ofmicroorganisms and conservative solute tracers (Harvey, 1997). Harvey (1997)described movement of microorganisms and solute tracers in relatively uniformgranular material, like sand, fractured media, and stratified granular material thatconsists of adjacent layers of finer and coarser sand. In relatively homogeneoussand, viruses are transported at the same velocity as a conservative salt tracer(Pieper et al., 1997; Schijven et al., 1999). However, in fractured media thekinetically or spatially available flow path can be considerably smaller for micro-organisms than for a solute tracer (Harvey, 1997). Many studies of bacterialmovement exist that generally show a rapid movement of bacteria due to prefer-ential flow through macropores, cracks, worm holes, and channels formed by plantroots (Abu-Ashour et al., 1994). Even more generally, colloids have been shownto be most mobile in soils with relatively large pores (Ouyang et al., 1996). Infractured or stratified soils, tracers like chloride or bromide are small enough toadvect or diffuse into fine fractures that are connected to preferential flow pathsthat are inaccessible to microorganisms (Harvey, 1997). Breakthrough of a solutein such heterogeneous soils will, therefore, be partially retarded and much moredispersed than a colloidal particle such as a virus. In addition, the solute tracer willfail to reach a steady state value for C/C0 of 1 as was shown by Bales et al. (1989).Breakthrough curves of solute tracers in heterogeneous soils will typically have theshape of curve E in Figure 3a. Diffusion into a matrix with small pores has the sameretarding effect as kinetically limited adsorption. Therefore, time to peak break-through of the solute will be later than that of the virus. In these cases, where theaverage flow velocity of viruses is higher than that of water, v in Equation 1 shouldbe interpreted as the flow velocity of the viruses. Rehman et al. (1999) developeda stochastic model for virus transport that accounts for effects of spatial variabilityin aquifer hydraulic conductivity. They could simulate that a high degree of aquiferheterogeneity can lead to virus breakthrough preceding that of a conservativetracer.

Bales et al. (1989) found that 35 to 40% of MS2 was excluded from the porevolume in a column with sandy soil. Bales et al. (1989) also studied transport of



105

phage f2 and two solute tracers, isothiocyanate and pentafluorobenzoic acid incolumns with nonsorbing fractured rock. The breakthrough curve of f2 showedlittle dispersion, whereas the solute tracers were dispersed, because of diffusioninto the porous matrix. The authors concluded that bacteriophages are better tracersfor transport of colloidal contaminants than solutes, because they were not subjectto matrix diffusion. McKay et al. (1993) showed that MS2 and PRD1 traveled atrates of 2 to more than 5 m/d in a fractured clay-rich till, whereas bromide and 18Owere transported at rates of only 0.01 to 0.07 m/d. Also, in a clay-rich till, Hinsbyet al. (1996) showed that MS2 and PRD1 were transported at high rates of 4 to 360m/day. These high rates were similar to water flow rates in the fractures. Thechloride concentration took more than 43 h to approach steady state, because ofdiffusion into the small pores in the matrix between the fractures. The phagesreached steady state in less than 3 h, because of pore size exclusion. Rossi (1994)showed that bacteriophages H40/1 and H6/1 were transported in karst faster thanuranine and naphtionate due to preferential flow. Rossi et al. (1994) showed thatbacteriophages T7 and f1 migrated about three times faster than naphtionate, basedon time to peak breakthrough. Migration rates of the tracers could be correlatedwith permeability distribution obtained by radio-magneto Tellury measurements.This showed that the tracers followed the more permeable pathways. The same wasfound with phages H40/1, H6/1, T7, and Psf2 compared with uranine by Rossi(1994) in a karstic aquifer.

Paul et al. (1995) used bacteriophages PRD1and ϕHSIC1 as tracers to inves-tigate fate and transport of sewage through a highly porous limestone matrix tomarine waters (Key Largo). Bacteriophage ϕHSIC was seeded into a septic tankand PRD1 at an injection point for disposal of domestic wastewater. Estimatedrates of migration of the two bacteriophages ranged from 14 to 580 m/day, whichis over 500-fold greater than the water velocity, which was measured by subsurfaceflow meters. The phages traveled at least a distance of 800 m. Concentrations ofthe viruses varied with the falling tides. However, in this study, it was not clearwhether the phages really traveled faster than the water because measurement ofwater flow was not carried out at the same time and place. Heavy rainfall precededthis study and may have caused unusually high flow rates. It was not clear whetherPRD1 could reach surface marine waters from the injection point. In a subsequentstudy, Paul et al. (1997) repeated seeding with bacteriophages MS2, PRD1, andϕHSIC1 at Key Largo, but also at another disposal well (Middle Keys). Thedifferent phages traveled at similar rates and made their way rapidly to surfacemarine waters from both injection wells, but transport rates were much greater inKey Largo than in the Middle Keys, 460 m/day and 24 m/day, respectively. Thisdifference in flow rates could be ascribed to differences in tidal pumping, and alsoto the fact that there exist manmade channels through the limestone at the KeyLargo site, but not at the Middle Keys site.

Sinton et al. (1997) studied the movement of somatic coliphages, FRNAPHsand fecal coliforms through an alluvial gravel aquifer where effluent was irrigated.



106

Breakthrough was followed at 60 m and about 400 m. In a second experiment,MS2, E. coli J6-2, and rhodamine WT dye were injected and could also be detectedat 400 m. In both experiments, the phages exhibited the shortest times to peakconcentration, followed by the bacteria, and then the dye. Sinton et al. (1997)interpreted these data by suggesting that the bacteria were transported faster thanthe dye because of pore-size exclusion; the phages were transported even fasterthan the bacteria, because the electrostatic repulsion of the phages is supposedlystronger than that of the bacteria, or that the phages are more often attached to otherparticles of a size that are transported faster than bacteria. Sinton et al. (1997) basedtransport velocity of the dye, the phages, and the bacteria on the arrival time ofpeak concentrations. As was explained in Section III.B, differences in time to peakconcentration can also be an effect of different values of kinetic attachment anddetachment rate coefficients (curves E and F in Figure 3a). Thus, it may reasonedthat the difference between katt and kdet was much less for the bacteria than for thephages, which is a more plausible explanation for the observed difference in timeto peak concentration between the bacteria and the phages. In addition, kineticeffects may have played some role in the breakthrough of the rhodamine WT dye.A study exists where it was shown that rhodamine WT does not behave conserva-tively, but exhibited nonequilibrium adsorption (Di Fazio and Vurro, 1994).

To conclude, advection and dispersion are properties of the soil and water, andare commonly determined by means of solute (salt) transport experiments. How-ever, viruses may travel faster than the average water flow and show a smallerdispersion than a solute. Several studies have shown that viruses are transportedfaster than classic solute tracers, because they can be excluded from small poresand therefore preferentially follow more permeable pathways. This shows that formovement of colloids, bacteriophages are better tracers than solutes.

VIII. MODEL VIRUSES

A. Introduction

As pointed out in Section I.D, model viruses are needed to represent thebehavior of pathogenic viruses during subsurface transport and to predict theirremoval. Bacteriophages MS2, PRD1, ϕX174, and naturally occurring FRNAPHshave been used extensively to study virus transport under various column and fieldconditions. In this section, the role of these bacteriophages as model viruses isevaluated.

B. MS2

MS2 is an icosahedral phage with a diameter of 27 nm and a low isoelectricpoint of 3.5. The three-dimensional structure of its capsid is known at the atomic



107

level (Penrod et al., 1996). MS2 may be considered as a relatively conservativetracer for virus transport in saturated sandy soils at pH 6 to 8 and with a low organiccarbon content, because under those conditions it showed little or no adsorption(Bales et al., 1989; Powelson et al., 1990; Herbold-Paschke et al., 1991; Kinoshitaet al., 1993; Jin et al., 1997; Schijven et al., 1999). A low organic content wouldimply low hydrophobicity of the soil.

In most soils, attachment of MS2 is also relatively low compared with mostother viruses (Goyal and Gerba, 1979). Herbold-Paschke et al. (1991) showed thatonly a few percent of bacteriophages T4, MS2, and ϕX174 attached in 1-mcolumns filled with coarse or medium-grade quartz sand. MS2 removal was lessthan or equal to that of T4 and ϕX174, and much less than that of simian rotavirusSA11. In columns with a sandy soil (pH 7.5), Bradford et al. (1993) showed thatMS2 was removed less (0 to 68%) than poliovirus 1 (99%) and simian rotavirusSA–11 (90%). Farrah and Preston (1993) showed less or equal removal of MS2compared with poliovirus 1, coxsackievirus B5, and echovirus 5 in column withsand that was or was not modified by precipitation of metallic salts. Bales et al.(1993) showed that the sticking efficiency of MS2 in columns with silica beads wastwo to three times lower than that of poliovirus 1 at pH 7, but about the same atpH 5 to 5.5 (Table 2). Sobsey et al. (1995) observed that removal of MS2 in 10-cmcolumns with clay loam (52% clay) at pH 6 to 8 was similar to that of HAV,poliovirus 1, and echovirus 1. Removal of MS2 in columns with organic muck wasless than or equal to that of these viruses. Penrod et al. (1996) and Redman et al.(1997) showed that due to steric repulsion, attachment of MS2 to silica was lessthan that of bacteriophage λ and recombinant Norwalk virus particles to quartzsand at pH 5 to 7, although both had a stronger negative surface charge than MS2.Jin et al. (1997) compared transport of MS2 and ϕX174 in short sand columns(10 or 20 cm). MS2 did not adsorb but ϕX174 was retained significantly. In thefield studies of DeBorde et al. (1999) and Schijven et al. (1999) removal of MS2and PRD1 was similar. In the same study of DeBorde et al. (1999) removal ofϕX174 was slightly more efficient and that of poliovirus 1 (CHAT) even more.

However, possibly due to its hydrophobicity, MS2 may attach more than otherviruses to some soils (e.g., to soil K in the batch study of Goyal and Gerba [1979].Using continuously circulating columns of a sandy soil (95% sand, 7% silt, 2%clay, pH 7.1), Dowd et al. (1998) found that the attached fraction of MS2 wasgreater than that of four other bacteriophages, PRD1, Qβ, ϕX174, and PM2,possibly due to the presence of multivalent cations. In the same study of Dowd etal. (1998), attachment of MS2 to the sandy soil in a flow-through column wasintermediate.

With regard to inactivation, MS2 is less stable than several pathogenic viruses(Table 5). MS2 is inactivated faster at higher temperatures, but at temperatureslower than 7°C, its inactivation rate is very low and similar to that of PRD1 (Yateset al., 1985; Yahya et al., 1993; Blanc and Nasser, 1996; Schijven et al., 1999).



108

In conclusion, MS2 meets the requirements for a worst-case model virus,provided the water temperature is less than about 10°C, the soil does not containtoo many hydrophobic sites, and the concentration of multivalent cations is low.

C. PRD1

PRD1 is an icosahedral bacteriophage with a diameter of 62 nm with an innerlipid membrane (Bales et al., 1991; Caldentey et al., 1990). Its pI lies between3 and 4 (Loveland et al., 1997). PRD1 may be considered as a worst-case modelvirus because of its low inactivation rate between 10 to 23°C (Yahya et al., 1993;Blanc and Nasser, 1996; see also Table 5). Because of its larger size, PRD1 is ofinterest as a representative of rotaviruses and adenoviruses (Sinton et al., 1997).With regard to attachment characteristics, PRD1 seems to behave less conserva-tively than MS2 (Bales et al., 1991; Kinoshita et al., 1993; Powelson et al., 1993;Dowd et al., 1998), possibly because it is more hydrophobic than MS2 (Shields andFarrah, 1987; Bales et al., 1991; Kinoshita et al., 1993; Lytle and Routson, 1995).However, in a sandy aquifer (Bales et al., 1997) PRD1 attached at a much lowerrate than compared with the column studies (using the same soil) by Kinoshita etal. (1993). Detachment was much slower than attachment, but the detachment rateincreased 104 to 105 times by a high pH-pulse (Bales et al., 1997). In the fieldstudies by DeBorde et al. (1999) and Schijven et al. (1999) removal of PRD1 wassimilar to that of MS2. Apparently, PRD1 may also be considered as a relativelyconservative model virus, similar to MS2, under field conditions in sandy soils atpH 6 to 8 and with low organic carbon content and low concentration of multiva-lent cations. In addition, PRD1 is more stable at higher temperatures (12 to 23°C).

D. ϕX174

ϕX174 is less hydrophobic than MS2 (Shields and Farrah, 1987). In studies onthe retention of viruses by barrier materials such as membranes, condoms, andtesting gloves, ϕX174 is regarded as the best model virus, because it exhibits theleast electrostatic and hydrophobic interaction (Shields and Farrah, 1987; Lytle andRoutson, 1995; Fujito and Lytle, 1996). ϕX174 essentially has no charge at neutralpH (pI = 6.6 to 6.8) and a size of 27 nm (Fujito and Lytle, 1996; Dowd et al., 1998).

Jin et al. (1997) found that breakthrough of ϕX174 in columns with Ottawasand attached significantly, whereas MS2 did not. It was suggested that thisdifference in adsorption behavior was a reflection of the difference in the isoelec-tric points of the two viruses. The authors suggested that because the pI of ϕX174is the same as that of poliovirus 1, it should have similar attachment behavior.However, Funderberg et al. (1981) already showed that these viruses behavedifferently. They studied removal of poliovirus 1, reovirus 3, and bacteriophage



109

ϕX174 in columns of eight different soils. The strongest positive correlation wasfound between removal of poliovirus 1 and reovirus 3, and soil cation exchangecapacity. Whereas removal of these enteroviruses was correlated more with soilproperties, removal of ϕX174 was affected most by the residence time in thecolumn. Removal of ϕX174 was usually less than that of poliovirus 1. In columnswith quartz sand, removal of ϕX174 was less than that of bacteriophage T4, andmore than or equal to that of MS2. However, removal of all these phages was quitelow (Herbold-Paschke et al., 1991). Dowd et al. (1998) showed that in a sandy soilin circulating-flow columns, and in flow-through columns removal of ϕX174 wasless than that of MS2, PRD1, PM2, and Qϕ. In field studies of DeBorde et al.(1998, 1999), ϕX174 appeared to be very stable (i.e., its inactivation was negligibleover a period of about one-half year).

ϕX174 may be a relatively conservative model virus because of its lowhydrophobicity (Shields and Farrah, 1987) and stability (DeBorde et al., 1998,1999). In soils, where hydrophobic interactions would significantly increase virusremoval, ϕX174 would be a better choice as a model virus than for instance, MS2or PRD1. However, the value of pH will strongly determine whether ϕX174 willbehave conservatively. A pH range of 6 to 8 is very common for most soils, andat pH 6 the net surface charge of ϕX174 will be positive, but at pH 8 it will benegative.

E. FRNAPHs

FRNAPHs have similar physical properties as enteroviruses, especially withrespect to size (Bitton, 1980; Havelaar, 1993). MS2 belongs to group I of FRNAPHs(Havelaar, 1986). As naturally present model viruses they are of high interest torepresent enteroviruses in various treatment processes of surface water, includingsoil passage. Before entering a treatment like soil passage, enteroviruses andFRNAPH largely have followed the same path (i.e., both have passed the seweragesystem, followed by sewage treatment, discharge into surface water, and some kindof pretreatment before recharge into soil). It may be reasoned that along this pathfrom the sewage system to the point of recharge into an aquifer, viruses that are lessstable, or that adsorb readily to solid surfaces, have disappeared already. Thissuggests that a selection has taken place of very stable and poorly adsorbing viruses(i.e., worst-case viruses). This selection has been the same for FRNAPHs andenteroviruses.

In surface water, FRNAPHs occur in numbers of 102 to 104 times higher thanenteroviruses (Havelaar et al., 1993). Therefore, it has been possible to show 4 to6 log units removal of FRNAPH’s by riverbank filtration (Havelaar et al., 1995).

In an alluvial gravel aquifer where effluent was irrigated, Sinton et al. (1997)studied transport of somatic coliphages, FRNAPH and faecal coliforms in oneexperiment and MS2 and E. coli J6-2 in a second experiment. Concentrations of



110

all these microorganisms were similarly reduced, about 9 log10 after 400 m oftransport. Concentration of the rhodamine WT dye was reduced about 7 log10 overthis distance. This shows that removal of FRNAPH and MS2, also an FRNAPH,is similar. Low sticking efficiencies of naturally present FRNAPH have been foundby Schijven et al. (1998) under field conditions (Table 3). DeBorde et al. (1998)found a removal rate of naturally present FRNAPHs and somatic coliphages thatwas half of that of bacteriophages MS2 and ϕX174. However, the latter twobacteriophages were seeded as slug injections, whereas FRNAPHs were continu-ously introduced with sewage effluent.

In the study of Nasser et al. (1993), FRNAPH appeared to be very persistentat 10 to 30°C, and not affected by the type of water. They were even morepersistent than HAV, which can be regarded as a very persistent virus (Sobsey etal., 1986; Nasser et al., 1993; Blanc and Nasser, 1996). However, Chung andSobsey (1992) observed that FRNAPHs are less stable than HAV in seawater.

In conclusion, FRNAPHs as a group of naturally occurring viruses are veryuseful model viruses for the behavior of viruses during subsurface transport.FRNAPHs behave relatively conservatively, like MS2, and they have been shownto be very persistent. Moreover, naturally present FRNAPHs may consist of stableand poorly adsorbing viruses prior to treatment by soil passage.

IX. VIRUS REMOVAL BY SOIL PASSAGE

A. Relative Contributions of Adsorption and Inactivation to VirusRemoval

Thus far in this article, adsorption and inactivation have been evaluated sepa-rately. This section focuses on the relative contributions of these processes to virusremoval by soil passage. Virus adsorption to soil is the most important process forattenuation. However, actual removal (i.e., disappearance of viruses), is due toinactivation, and also of irreversible attachment, if it exists. Under steady-stateconditions, the relative contributions of inactivation and adsorption to the removalof viruses by soil passage can be compared easily. A steady-state situation occurswhen input of virus is continuous. This usually is the case (e.g., during bankfiltration, dune recharge, deep well injection, or continuously leaking sewagepipes) and may be seen as a worst-case situation. Considering a one-dimensionalsteady-state situation, Equation 1 may be simplified to:

α ∂∂

∂∂L

C

x

C

x

Q

nv

2

2 – = (29)

Here, αL is the longitudinal dispersivity, [L]. Under steady-state conditions, itfollows from Equation 3 that:



111

µ =s kin B kin att B kinS nk C k S, det–ρ ρ (30)

Substitution of Equations 2 and 31 into 4 gives:

Q nCn

kk

kl s eqB

eqatt s kin

s kin

= µ + µ +µ+ µ

λ λρ

,,

det ,

where = (31)

Now, the solution of Equation 29 can be written as:

log.

–C

C

x vL

L0 2 3

1 1 4

2

=+

α λ

α(32)

Where C0 is the concentration at x = 0, and log(C/C0) is defined as virus removal.From Equations 31 and 32, the relative contributions of adsorption and inactivationto virus removal can be deduced. Because equilibrium adsorption has been shownto be of little importance in several studies (Bales et al., 1991, 1993, 1997; Pieperet al., 1997; Jin et al., 1997; Schijven et al., 1999), the second term in Equation 31may be neglected. From Equation 31 it can be seen that if there is no inactivationat all, there will be no removal (λ = 0) under steady-state conditions. If there is nodetachment, removal will be determined by inactivation and irreversible attach-ment of free viruses (λ = µl + katt).

In the field study by Schijven et al. (1999), it appeared that kdet << µs,kin and katt

> µl; therefore, λ ≈ katt. In that case, virus removal is mainly determined byattachment, which appears to work as irreversible attachment. In a field study ofBales et al. (1997), virus inactivation was considered to be insignificant. Theauthors concluded that virus removal could be described primarily by katt and kdet,and that simply knowing a retardation factor is not sufficient. This is justified onlyif katt is much smaller than µs. Also, Pieper et al. (1997) and DeBorde et al. (1999)regarded virus inactivation as negligible during their field experiments, and con-sidered virus removal to be solely determined by attachment. Although detachmentrates were not calculated in these studies, the shapes of their breakthrough curvessuggest that detachment rates were much lower than attachment rates.

Table 8 summarizes values of katt, kdet, µl, and µs for MS2 and PRD1 calculatedfrom observed virus removal in a few field studies. Values for katt and kdet have beenreported in only two field studies. Bales et al. (1997) reported two values for each(Table 8). During a high pH pulse, katt was only a factor 2 lower, whereas kdet

increased by a factor of 5 × 104. In that case kdet was larger than katt and probablymuch larger than µs implying that the kinetic term in Equation 31 took the form ofthe adsorption equilibrium. Schijven et al. (1999) found different katt and kdet valuesfor different travel distances (see next section).



112

TABLE 8Values of �, katt, kdet, �l and �s, [day –1], for MS2 and PRD1 from FieldStudies

Parameter MS2 PRD1 Ref.

katt 6.1 – 11 Bales et al. (1997)0.8 – 4.1 0.7 – 4.0 Schijven et al. (1999)

kdet 0.0003 – 15 Bales et al. (1997)0.00087 – 0.0030 0.00077 – 0.0034 Schijven et al. (1999)

µl (5 ± 3 °C) 0.03 0.12 Schijven et al. (1999)µs (2 — 5 °C) 0.09 0.07 Schijven et al. (1999)

All in all, it appears that only very few data are available on attachment anddetachment of viruses under field conditions. In order to be able to predict virusremoval, more values of attachment, detachment, and inactivation rate coefficientsfor viruses under various conditions are needed. Comparison of the α-values fromTables 2 and 3 give the impression that, at least for short travel distances, thesevalues are similar for column and field situations. However, this needs verification.Except during perturbations like a high pH, it seems that kdet is generally muchsmaller than katt, which is consistent with observations from several column studies(Bales et al., 1991, 1993; Kinoshita et al., 1993; Penrod et al., 1996; Dowd et al.,1998). Possibly, the ratio of katt/kdet from column scale may be extrapolated to fieldscale under similar conditions. However, this needs to be verified too. The inac-tivation rate coefficients that were found in a field study by Schijven et al. (1999)are similar in value to those from the batch studies that are summarized in Tables5 and 6. This suggests that the values for inactivation rate coefficients as found inbatch experiments may be used for prediction of virus transport at field scale.However, more information is needed to verify this.

Yates et al. (1986) calculated safe setback distances, solely based on inactiva-tion rates of MS2. To that purpose, they sampled groundwater from 71 sites withina certain area and determined inactivation rates of MS2 in these samples at thecorresponding in situ temperatures (20 to 30.5°C). Virus inactivation rates of othernearby sites in the area were calculated by means of kriging, a geostatisticaltechnique. A contour map could be made that showed the variation in separationdistances assuring 7-log10 reduction in virus concentrations by inactivation. Thisapproach was extended by combining kriging with a linear relation for µl of MS2with temperature. This eliminated the need of having to measure virus inactivationrates in the laboratory (Yates and Yates, 1987). This way, similar maps of setbackdistances could be constructed. Yates and Yates (1988) extended this model evenfurther to account for alterations in the flow field by pumping. This geostatisticalmodel may, however, underestimate setback distances, because at higher tempera-tures, the inactivation rate of MS2 is higher than that of other viruses, such as HAV



113

(Yates et al., 1986; see also Section VI.B). On the other hand, setback distancesmay be overestimated grossly this way, because the contribution of inactivation ofattached viruses was not included (see Equation 31).

B. Removal of Viruses with Distance

From Equation 32, with constant rate coefficients for attachment, detachment,and inactivation, it can be deduced that in a saturated soil under steady-stateconditions, virus removal (log10(C/C0)) should decline in a linear fashion withtravel distance. In several field studies (Bales et al., 1995; Pieper et al., 1997;DeBorde et al., 1998, 1999) viruses were seeded as slug injections, thus no steadystate was achieved. In that case, apparent removal rates are expected to increasewith distance due to dispersion. However, several column and field studies haveshown that the removal rate appears to be initially higher.

Gerba and Lance (1978) have shown that approximately one log10 of poliovirus1 was removed (C/C0 = 0.1) from primary sewage during passage through the first5 cm of a column with loamy sand. An additional 35-cm travel distance wasnecessary to reduce the virus concentration another log10. Similar results wereobtained when the column was flooded with secondary sewage. Wang et al. (1981)analyzed removal of poliovirus 1 and echovirus 1 after continuous application tocolumns with three different sands and a sandy loam for 3 to 4 days at constant flowrates. Removal of both viruses appeared to be very similar. Statistical analysisindicated that the rate of virus removal in the upper 17-cm depth of the soil columnwas significantly greater than in the lower depths of the soil column. In a fieldstudy of Bales et al. (1995), concentration of PRD1 was attenuated about 5 log10

after the first 2 m of passage in a sandy aquifer. Attenuation of PRD1 was estimatedrelative to that of bromide to account for dilution effects. However, an additionalattenuation of only one log10 during the following 9 m was observed. In anotherfield study of Bales et al. (1997), bacteriophages PRD1 and M1 both were removedabout 4 log10 after 0.94 m of passage through a sandy aquifer, but only another onelog10 after the following 1.6 m. A conservative tracer did not decrease similarlyand, therefore, dispersion was ruled out as a possible cause. Pieper et al. (1997)presented RB of PRD1 with distance at a field site with a sewage-contaminated andan uncontaminated zone. They found that the strongest attenuation occurred duringthe first 1.8 m, but concentrations of PRD1 remained nearly constant during thenext 1.8 m. Plotting RB of PRD1 on a log scale showed that the removal rate ofPRD1 was initially higher. Similar results were obtained by Schijven et al. (1998).They measured efficient reduction of FRNAPHs by 3.8 log10 after 2 m of duneinfiltration. However, at 4 m, an additional reduction of only 0.83 log10 was found,albeit with a large uncertainty. Consequently, the corresponding estimated value ofthe sticking efficiency is much lower at 4 m than 2 m (see Table 3). Schijven etal. (1999) seeded bacteriophages MS2 and PRD1 during 11 days at a field site for



114

dune recharge. Removal of MS2 and PRD1 was very similar. Bacteriophageconcentrations were reduced about 3 log10 within the first 2.4 m and another5 log10 in a linear fashion within the following 27 m. DeBorde et al. (1999) reportedremoval with distance of MS2, PRD1, ϕX174, and poliovirus 1 (CHAT) underfield conditions. Maximum breakthrough concentrations of MS2, PRD1, andϕX174 decreased about 2.5 log10 with 8 m and an additional 3.5 to 5 log10 withinthe next 32 m. Concentrations of poliovirus 1 decreased 4 log10 within the first 8 mand about 1 log10 in the following 12 m. In another field study by DeBorde et al.(1998) removal rates of MS2 and ϕX174 seemed to decrease with distance, but notsignificantly. MS2 and ϕX174 were seeded as slug injections. In other words, if theviruses had been seeded for a longer period of time, such that semi-steady-stateconditions could have been assumed, the removal rate would likely have declinedwith distance. However, removal rates of somatic and FRNAPHs were found to belinear over a distance of 18 m. These naturally occurring bacteriophages werepresent due to continuous infiltration of sewage effluent.

With the exception of FRNAPHs and somatic coliphages in the study byDeBorde et al. (1998), these studies all suggest that initial higher removal ofviruses with distance is typical. This phenomenon may be explained by soil and/orvirus heterogeneity.

Soil heterogeneity may be a cause, that is if more sites of attachment areavailable in the first centimeters of passage through a column or in the first meterat a field location. Although this may be the case in field studies, in the columnstudies that are referred to here, the columns were filled in small increments toobtain uniform packing of the soil. Therefore, it seems unlikely that the firstcentimeters contained a larger proportion of silt and clays, or smaller grains thatcould cause more efficient attachment. The field studies of Bales et al. (1995) andPieper et al. (1997) were performed in the same aquifer that was contaminated bysewage disposal. The soil consisted of stratified, well-sorted, medium to coarsesand with some gravel, and groundwater velocity ranged from 0.2 to 0.7 m/day andporosity from 20 to 40% (Bales et al., 1995). From these studies, it is not clear ifinjection of the studied bacteriophages took place in finer soil material. Further-more, nonlinear removal with distance appeared to be independent of the extent ofsewage contamination (Pieper et al., 1997). From the field study by Bales et al.(1997), it is not clear whether PRD1 and M1 were injected at a point with finer soilmaterial. Extensive soil analysis in the study by Schijven et al. (1999) showed thatthe average grain size of the soil was even larger in soil samples from the first 5 and10 to 15 cm than in soil samples taken at the monitoring wells. Therefore, thepossibility that fine-grained sediments at the bottom of the compartment were thecause of high early attachment could be ruled out. On the other hand, higherconcentrations of bonded organic matter were found in the soil samples at a shorterdistance. This suggests that the bacteriophages may have been removed moreefficiently during the first meters of soil passage by hydrophobic interactions.



115

However, from the above, it appears that soil heterogeneity does not satisfactorilyexplain initial higher virus removal.

Viral heterogeneity may be a cause for the greater initial removal (i.e., theviruses in the suspensions that were used for seeding a column or a field site mayhave different affinities for attachment sites), as was suggested by Pieper et al.(1997) and Schijven et al. (1999). Figure 14 shows how sticking efficiency α forMS2 and PRD1 gradually decrease with distance, as was observed in a field studyby Schijven et al. (1999). Albinger et al. (1994) reported a range of stickingefficiencies for bacteria, even in the extreme case of a uniform collector surfaceand a monoclonal bacterial population. These variations in α could neither beexplained by preferential retention of larger cells nor by intrapopulation geneticvariation. Although the outer surface of a virus particle is less complex than thatof a bacterial cell, virus particles that are released from their host cell may be stillattached to cell debris, which would account for differences in adsorption betweenvirus particles. Furthermore, Goyal and Gerba (1979) have shown that virusadsorption to soil not only depends on the type and strain of virus under consid-eration, but may also vary between isolates of the same virus type. Viruses may bepresent as aggregates (see Section V.A), and it has been shown that viruses attachreadily to other colloidal particles, like clays that are ubiquitously present insewage water and surface water (see Section VI.C). It may be reasoned thatcombinations of viruses and other colloidal particles thus behave as particles withdifferent size and density and therefore different collision efficiencies, but alsowith different sticking efficiencies. In addition, different types of viruses that are

FIGURE 14. Sticking efficiency α for MS2 and PRD1 with travel distance (Schijven et al.,1999).



116

attached to the same type of colloidal particles may show the same behavior duringsubsurface transport. Possibly, transport of a fraction of virus particles may beenhanced, due to their attachment to inorganic colloids such as silicate clays andiron or aluminium oxide clays. Such colloidal particles are released in an anoxicenvironment and can migrate considerable distances (Ouyang et al., 1996).

Viral heterogeneity is the most likely explanation for the nonlinear removal ofFRNAPHs by dune recharge (Schijven et al., 1998). Although FRNAPHs form arather homogeneous group, they consist of different virus types that differ in sizeand surface charge. Therefore, they are likely to differ in sticking efficiencies. Itcan be argued that FRNAPHs with higher αs attach faster, thereby lowering theaverage α of the population of free phages, as they are transported further.

Rehman et al. (1999) developed a large-scale model of virus transport inaquifers using spectral perturbation analysis. A sensitivity analysis showed thataquifer heterogeneity can lead to virus breakthrough actually preceding that of aconservative tracer. Also, use of a heterogeneous colloid filtration term results inhigher predicted concentrations than the use of a simple first-order constant attach-ment rate. It was pointed out that measurements of the spatial variability of thesticking efficiency would be extremely important in predicting virus transport.

In conclusion, removal of viruses by soil passage apparently declines withdistance. This has important consequences for prediction of their removal, thusalso for the calculation of setback distances to adequately protect groundwatersources and to assure adequate treatment of infiltrated surface water. From theabove, it is clear that predictions of virus removal at larger distances are severelyoverestimated if they are based on removal data from column experiments or fromfield studies where transport was studied over short distances.

To improve predictions on the removal of viruses by soil passage, knowledgeof the soil heterogeneities at the location of interest is clearly needed. Also,heterogeneity of the transported virus particles (i.e., the distribution of stickingefficiencies within the population of virus-particles), including its cause should beinvestigated further.

X. SUMMARY AND CONCLUSIONS

A. Adsorption

Many batch experiments with suspensions of viruses and soil have beencarried out to investigate the effects of various factors on virus–soil interactions.Commonly, on a time scale of only a few hours, virus inactivation is negligible andequilibrium adsorption has been reported to be reached, or has been assumed to bereached. Also, in many column and field studies, adsorption of viruses duringtransport was assumed to reach equilibrium. This is described by retardationcoefficient R, whereby a value of R greater than one reflects retarded breakthrough



117

due to equilibrium adsorption. However, only in some cases, retardation coeffi-cients of about 2 to 5 have been reported. Usually little or no retardation is found.Therefore, retarded breakthrough by equilibrium adsorption is believed to be oflittle significance. Kinetically limited attachment and detachment mainly governremoval of viruses during transport and determine the shape of breakthroughcurves. In field studies, it appeared that attachment rates were relatively fast,whereas detachment rates may be much slower. Dependent on soil heterogeneity,viruses may travel faster than the average water flow and show a smaller dispersionthan a solute, because they can be excluded from small pores and, therefore,preferentially follow more permeable pathways.

In a batch suspension of viruses and soil, adsorption equilibrium is not reachedinstantaneously. Instead, virus attachment at the microscale can be described as theresult of mass transport, within a pore, to the solid surface and subsequent immo-bilization at the surface by physical and possibly chemical interactions. The overallrate of attachment depends on which of these two processes is the rate-limitingstep. At pH 7 to 8, as in many aquifers, the net surface charge of most viruses andsoils is negative, and thus conditions for attachment are generally unfavorable.Under such conditions, virus–surface interaction is the rate-limiting process forattachment. In the absence of repulsive energy barriers, or in the presence ofattractive double-layer interactions, conditions for attachment are favorable. In thatcase, mass transport to the vicinity of the soil surfaces is the rate-limiting step.Favorable chemical conditions for attachment may develop in groundwaters withhigh levels of water hardness and ionic strength. Attachment will also be favorablefor solid surfaces (or patches on solid surfaces) that are positively charged due toiron, aluminium, or manganese oxide coatings. Attachment to such favorablepatches may be irreversible. Several studies support the concept that virusespreferably attach only to a fraction of the soil surface having favorable chargecharacteristics.

Unfortunately, estimates of adsorption parameters from batch experimentsappear to be of little use in predicting adsorption of viruses in column or fieldexperiments. Values for adsorption obtained from different batch studies are evendifficult to compare because there is no standard protocol. They are highly variabledue to heterogeneity of soil preparations, different sizes and types of containers,different methods of agitation, and differences in the air–water interface. Mostbatch studies were carried out to measure equilibrium adsorption, whereby attach-ment and detachment rate coefficients are not determined. Only one batch studyexists where the kinetic part (that is operative before equilibrium is reached) wasanalyzed (Grant et al., 1993). Therefore, values of attachment and detachment ratecoefficients from batch studies are not available, but even if they were, they are notapplicable for predicting virus removal in transport situations. Due to the stirringin a batch experiment the number of accessible sites for attachment is much higherthan in a column. In a column, attachment rates are therefore much lower. Theopposite is probably true for the detachment rate. The detachment rate coefficient



118

for a virus in a batch system is probably smaller than under transport conditions,where there is advective flow. This implies that the observed ratio katt/kdet from abatch experiment is expected to be much larger than that from a column or fieldexperiment. This difference between batch and column experiments has beenfound in several studies, where equilibrium adsorption was assumed.

A problem with many batch experiments reported in the literature is that themeasurements are probably stopped too early. Almost all batch experiments areused to evaluate equilibrium adsorption coefficients. However, in most cases, it isnot clear if equilibrium has been reached. The ratio of katt/kdet may therefore havebeen strongly underestimated. In several column and field studies at pH 7 to 8,where kinetically limited adsorption was considered, the ratio katt/kdet was found tobe very large. This is totally opposite to the expectation that katt/kdet from a batchexperiment is expected to be larger than that from a column or field experiment.Calculations predicted that equilibrium in a batch system would be reached onlyafter about 40 h. Also, this would mean that a low percentage of adsorption foundin a batch experiment after a short period of time is rather a consequence of slowadsorption than of equilibrium.

The major factor that affects adsorption is pH. At higher pH, electrostaticrepulsion increases, resulting in a decreased attachment rate and an increaseddetachment rate. In most aquifers, surface characteristics of the soil are heteroge-neous and also different viruses with different pIs may be present. Therefore,dependent on pH and thus on the charge of the virus and soil particles, adsorptionof some of these viruses may be irreversible, whereas that of others may bereversible. At pH 7 to 8, adsorption will mainly be reversible.

Dissolved organic matter may decrease virus attachment to soil because ofcompetition for the same binding sites. Dissolved organic matter such as surfac-tants may disrupt hydrophobic bonds between soil and virus, resulting in anincreased detachment rate. At the same time, viruses and many organic materialscontain hydrophobic groups on their surfaces. Therefore, once adsorbed, bondedorganic matter may provide hydrophobic binding sites for viruses. The presence ofsolid organic matter also may result in an increased attachment rate throughhydrophobic binding. The enhancing and attenuating effects of organic matter onadsorption of viruses and their dependence on soil and virus properties make theirquantification very difficult, especially under field conditions. Effects of organicmatter will therefore be responsible for considerable uncertainty in predicting virusremoval.

B. Inactivation

Commonly, virus inactivation is considered as a first-order rate process. Inac-tivation rates of viruses that are attached to solids may be reduced or enhanced. Inthis article, values of the inactivation rate coefficient µs for attached viruses were



119

calculated using data from some batch studies. It appeared that enhanced orreduced inactivation is very much virus-specific and almost independent of theratio katt/kdet. Reduced inactivation may be the result of protection from proteolyticenzymes or other virus-inactivating substances, increased stability of the viruscapsid when attached, prevention from aggregate formation, and blocking fromultraviolet radiation. Attachment of viruses to solids may also significantly reduceinactivation by preventing them from entering into the air–water interface, or theair–water–solid contact line. Enhanced virus inactivation in the presence of soilmay also occur in some cases, probably due to repeated attachment and detachmentcycles.

Temperature is the most important factor that influences virus inactivation.Inactivation rates increase with temperature. The temperature sensitivity of µl

depends on the type of virus. Probably, µs values change similarly with temperatureas µl values. Inactivation of some pathogenic viruses, such as HAV, is veryinsensitive to temperature changes. Microbial activity may increase inactivationrates of viruses under aerobic conditions.

C. Model Viruses

Adsorption and inactivation are strongly virus dependent. One way to modelthe removal of pathogenic viruses by soil passage is to use a cocktail of viruses thatrepresent a range of these characteristics. Alternatively, a virus that adsorbs lessand is more stable than other viruses under certain conditions may be consideredas a worst-case model virus.

MS2 meets the requirements of a worst-case model virus, provided the watertemperature is less than about 10°C and the soil does not contain too manyhydrophobic sites. PRD1 may also be considered as a relatively conservativemodel virus under field conditions in sandy soils at pH 6 to 8 and with low organiccarbon content. In addition, PRD1 is more stable than MS2 at higher temperatures.

Bacteriophage ϕX174 may be a relatively conservative model virus because ofits low hydrophobicity and stability. Thus, in soils with a high organic carboncontent, ϕX174 would be a better choice as a model virus than for example MS2or PRD1. However, the pH value will strongly determine whether ϕX174 willbehave conservatively. There is a significant change in its adsorption behaviorwhen pH varies from 6 to 8, a very common pH range for most soils. At pH 6 thenet surface charge of ϕX174 will be positive (high attachment), but at pH 8 it willbe negative (low attachment).

FRNAPHs, as a group of naturally occurring viruses, are very useful modelviruses. They behave relatively conservatively, like MS2, also an FRNAPH, andthey have been shown to be very persistent. Moreover, FRNAPHs that are presentin surface water or treated wastewater that is used for recharging groundwater,consist of stable and poorly adsorbing viruses.



120

In practice, only few viruses have been used as model viruses. Among them,MS2 seems to meet the requirements for a model virus very well. Nevertheless,when carrying out a field experiment, it is advisable to make use of two or threemodel viruses that span a range of properties, such as size, surface charge, andhydrophobicity. In that regard MS2, PRD1, and ϕX174 make a promising cocktail.

D. Virus Removal

In a recent field study, it was found that virus removal was determined mainlyby attachment (Schijven et al., 1999). Similarly, in other field studies, virusinactivation was considered to be insignificant, and therefore virus removal wasconsidered to be solely determined by attachment (Bales et al., 1995, 1997; Pieperet al., 1997; DeBorde et al., 1998, 1999). Although detachment rates were notcalculated in these studies, the shapes of their breakthrough curves suggest thatdetachment rates were much lower than attachment rates. Although large numbersof laboratory studies have been performed to investigate the factors affecting virustransport through the subsurface, only few well-defined field studies have beencarried out. Therefore, only very few data are available on attachment and detach-ment of viruses under field conditions. To be able to quantify virus removal undervarious field conditions, more values of attachment, detachment, and inactivationrate coefficients are needed. DeBorde et al. (1998) proposed to generate a databaseof a limited number of field studies that span the range of hydrogeological settings,conditions, and viruses. This article may be considered as a first step toward sucha database because it provides an inventory of field studies and parameter valuesobtained from these studies, even though these data are incomplete. An inventoryalso has been made of sticking efficiencies, values of attachment, detachment, andinactivation rate coefficients for attached and free viruses both at laboratory andfield scale.

Field experiments are costly, complex, and often raise more questions. It maytherefore still be useful and desirable to carry out column and batch experimentsto obtain parameters that can be extrapolated to the field situation. Possibly, valuesfor sticking efficiencies as found in column experiments are similar to those fromfield experiments, at least for short travel distances. The ratio of katt/kdet fromcolumn-scale possibly may be extrapolated to field-scale under similar conditions.The values for inactivation rate coefficients as found in batch experiments may beused for prediction of virus transport at field scale. The possibility of extrapolatingparameter values from laboratory scale to field scale needs to be verified. Further-more, laboratory scale experiments are needed to compare removal of modelviruses with that of pathogenic viruses.

Removal of viruses by soil passage appears to decline with distance. Possiblecauses for this nonlinear removal may be heterogeneities within the soil as well aswithin the population of transported virus particles. In the aqueous environment,



121

viruses appear to be partially attached to other colloidal particles. This may explainheterogeneity of the adsorptive characteristics of transported virus particles. Thenonlinearity of removal with distance has important consequences for predictionof virus removal, thus also for the calculation of setback distances that are neededto adequately protect groundwater sources and to ensure adequate treatment ofinfiltrated surface water. Predictions of virus removal at larger distances areseverely overestimated if they are based on removal data from column experimentsor from small-scale field studies. To improve predictions on the removal of virusesby soil passage, knowledge of the soil heterogeneities at the location of interest isneeded. Also, heterogeneity of the transported virus particles (i.e., the distributionof sticking efficiencies within the population of virus-particles), including itscause, should be investigated further.

ACKNOWLEDGMENTS

A. H. Havelaar, A. M. de Roda Husman, and E. J. T. M. Leenen from theNational Institute of Public Health and the Environment, Bilthoven, The Nether-lands are greatly acknowledged for their stimulating and expert comments.

REFERENCES

Abu-Ashour, J., Joy, D. M., Lee, H., Whiteley, H. R., and Zelin S. 1994. Transport of microorganismsthrough soil. Water Air Soil Pollut. 75, 141–158.

Albinger, O., Biesemeyer, B. K., Arnold, R. G., and Logan, B. E. 1994. Effect of bacterial hetero-geneity on adhesion to uniform collectors by monoclonal populations. FEMS Microbiol. Lett.124, 321–326.

Armon, R. and Cabelli, V. J. 1988. Phage f2 desorption from clay in estuarine water using nonionicdetergents, beef extract and chaotropic agents. Can. J. Microbiol. 34, 1022–1024.

Babich, H. and Stotzky, G. 1980. Reductions in inactivation rates of bacteriophages by clay mineralsin lake water. Water Res. 14, 185–187.

Bales, R. C., Gerba, C. P., Grondin, G. H., and Jensen, S. L. 1989. Bacteriophage transport in sandysoil and fractured tuff. Appl. Environ. Microbiol. 55, 2061–2067.

Bales, R. C., Hinkle, S. R., Kroeger, T. W., and Stocking, K. 1991. Bacteriophage adsorption duringtransport through porous media: chemical perturbations and reversibility. Environ. Sci. Technol.25, 2088–2095.

Bales, R. C., Li, S., Maguire, K. M., Yahya, M. T., and Gerba, C. P. 1993. MS2 and poliovirustransport in porous media: hydrophobic effects and chemical perturbations. Water ResourcesRes. 29, 957–963.

Bales, R. C., Li, S., Maguire, K. M., Yahya, M. T., Gerba, C. P., and Harvey, R. W. 1995. Virus andbacteria transport in a sandy aquifer, Cape Cod, MA. Ground Water 33, 653–661.

Bales, R. C., Li, S., Yeh, T. C. J., Lenczewski, M. E., and Gerba, C. P. 1997. Bacteriophage andmicrosphere transport in saturated porous media: forced-gradient experiment at Borden, Ontario.Water Resour. Res. 33, 639–648.

Bitton, G. and Mitchell, R. 1974. Effects of colloids on the survival of bacteriophages in seawater.Water Res. 10, 323–327.



122

Bitton, G. 1980. Introduction to Environmental Virology, pp. 1–28. New York, John Wiley & Sons.Bitton, G., Farrah, S. R., Ruskin, R. H., Butner, J., and Chou, Y. J. 1983. Survival of pathogenic and

indicator organisms in groundwater. Ground Water 21, 405–410.Blanc, R. and Nasser, A. 1996. Effect of effluent quality and temperature on the persistence of viruses

in soil. Water Sci. Technol. 33, 237–242.Bradford, S. M., Bradford, A. W., and Gerba, C. P. 1993. Virus transport through saturated soils.

Wiener Mitteilungen Wien 12, 143–147.Brush, C. F., Ghiorse, W. C., Anguish, L. J., Parlange J. Y., and Grimes, H. G. 1999. Transport of

Cryptosporidium parvum oocysts through saturated columns. J. Environ. Qual. 28, 809–815.Burge, W. D. and Enkiri, N. K. 1978.Virus adsorption to five soils. J. Environ. Qual. 7, 73–76.Caldentey, J., Bamford, J. K. H., and Bamford, D. H. 1990. Structure and assembly of bacteriophage

PRD1, an Escherichia coli virus with a membrane. J. Struct. Biol. 104, 44–51.Chung, H. and Sobsey, M. D. 1992. Survival of F-specific coliphages, Bacteroides fragilis phages,

hepatitis A virus (HAV) and poliovirus 1 in seawater and sediment. Water Sci. Technol. 27, 425–428.

Chrysikopoulos, C.V. and Sim, Y. 1996. One-dimensional virus transport in homogeneous porousmedia with time-dependent distribution coefficient. J. Hydrol. 185, 199–219.

DeBorde, D. C., Woessner, W. W., Lauerman, B., and Ball, P. N. 1998. Virus occurrence in a schoolseptic system and unconfined aquifer. Ground Water 36, 825–834.

DeBorde, D. C., Woessner, W. W., Kiley, Q. T., and Ball, P. N. 1999. Rapid transport of viruses ina floodplain aquifer. Water Res. 33, 2229–2238.

Di Fazio, R. and Vurro, M. 1994. Experimental test using rhodamine WT as tracer. Adv. WaterResource 17, 375–378.

Dizer, H., Nasser, A., and Lopez, J. M. 1984. Penetration of different human pathogenic viruses intosand columns percolated with distilled water, groundwater or wastewater. Appl. Environ.Microbiol. 47, 409–415.

Dowd, S. E., Pillai, S. D., Wang, S., and Corapcioglu, M. Y. 1998. Delineating the specific influenceof virus isoelectric point and size on virus adsorption and transport through sandy soils. Appl.Environ. Microbiol. 64, 405–410.

Farrah, S. R. 1982. Chemical factors influencing adsorption of bacteriophage MS2 to membranefilters. Appl. Environ. Microbiol. 43, 659–663.

Farrah, S. R. and Preston, D .R. 1993. Adsorption of viruses to sand modified by in situ precipitationof metallic salts. Wiener Mitteilungen Wien 12, 25–29.

Farrah, S. R., Preston, D. R., Toranzos, G. A., Girard, M., Erdos, G. A., and Vasuhdivan, V. 1991.Use of modified diatomaceous earth for removal and recovery of viruses in water. Appl.Environ. Microbiol. 57, 2502–2506.

Farrah, S. R., Shah, D. O., and Ingram, L. O. 1981. Effects of chaotropic and antichaotropic agentson elution of poliovirus on membrane filters. Proc. Natl. Acad. Sci. U.S.A. 78, 1229–1232.

Formentin, K., Rossi, P., and Aragno, M. 1997. Determination of bacteriophage migration andsurvival potential in karstic groundwaters using batch agitated experiments and mineral colloi-dal particles. Tracer Hydrol. 97, 39–46.

Fujito, B. T. and Lytle, C. D. 1996. Elution of viruses by ionic and nonionic surfactants. Appl.Environ. Microbiol. 62, 3470–3473.

Funderburg, S. W., Moore, B. E., Sagik, B. E., and Sorber, C. A. 1981. Viral transport through soilcolumns under conditions of saturated flow. Water Res. 15, 703–711.

Gantzer, C., Quignon, F., and Schwartzbrod, L. 1994. Poliovirus–1 adsorption onto and desorptionfrom montmorillonite in seawater, survival of the adsorbed virus. Environ. Technol. 15, 271–278.

Gerba, C.P. and Schaiberger, G.E. 1975. Effect of particulates on virus survival in seawater. J. WaterPollut. Control Fed. 47, 93–103.



123

Gerba, C. P., Stagg, C. H., and Abadie, M. G. 1978. Characterisation of sewage solid-associatedviruses and behaviour in natural waters. Water Res. 12, 805–812.

Gerba, C. P. and Lance, J. C. 1978. Poliovirus removal from primary and secondary sewage effluentby soil filtration. Appl. Environ. Microbiol. 36, 247–51.

Gerba, C. P., Goyal, S. M., Cech, I., and Bogdan, G. F. 1981. Quantitative assessment of theadsorptive behaviour of viruses to soils. Environ. Sci. Technol. 15, 940–944.

Gerba, C. P. 1984. Applied and theoretical aspects of virus adsorption to surfaces. Adv. Appl.Microbiol. 30, 133–168.

Gerba, C. P., Keswick, B. H., Dupont, H. L., and Fields, H. A. 1984. Isolation of rotavirus fromdrinking water. Monogr. Virol. 15, 119–125

Goyal, S. M. and Gerba, C. P. 1979. Comparative adsorption of human enteroviruses, simianrotavirus, and selected bacteriophages to soils. Appl. Environ. Microbiol. 38, 241–247.

Grant, S. B. 1994. Virus coagulation in aqueous environments. Environ. Sci. Technol. 28, 928–933.Grant, S. B. 1995. Inactivation kinetics of viral aggregates. J. Environ. Eng. 121, 311–319.Grant, S. B., List, E. J., and Lidstrom, M. E. 1993. Kinetic analysis of virus adsorption and

inactivation in batch experiments. Water Resour. Res. 29, 2067–2085.Hancock, C. M., Rose, J. B., and Callahan, M. 1998. Crypto and Giardia in US groundwater. J.

AWWA 90, 58–61.Harter, T., Wagner, S., and Atwill, E. R. 1999. Colloid transport and filtration of Cryptosporidium

parvum in sandy soils and aquifer sediments. Environ. Sci. Technol.Harvey, R. W. 1997. Microorganisms as tracers in groundwater injection and recovery experiments:

a review. FEMS Microbiol. Rev. 20, 461–472.Havelaar, A. H. 1986. F-specific bacteriophages as model viruses in water treatment processes. Ph.D.

Thesis, University of Utrecht, The Netherlands, 240 pp.Havelaar, A.H. 1993. Bacteriophages as models of enteric viruses in the environment. ASM News 59,

614–619.Havelaar, A.H., van Olphen, M, and Drost, Y.C. 1993. F-specific RNA bacteriophages are adequate

model organisms for enteric viruses in fresh water. Appl. Environ. Microbiolol. 59, 2956–2962.Havelaar, A. H., van Olphen, M., and Schijven, J. F. 1995. Removal and inactivation of viruses by

drinking water treatment processes under full scale conditions. Water Sci. Technol. 31, 55–62.Herbold-Paschke, K., Straub, U., Hahn, T., Teutsch, G., and Botzenhart, K. 1991. Behaviour of

pathogenic bacteria, phages and viruses in groundwater during transport and adsorption. WaterSci. Technol. 24, 301–304.

Hejkal, T. W., Wellings, F. M., Lewis, A. L., and LaRock, P. A. 1981. Distribution of virusesassociated with particles in wastewater. Appl. Environ. Microbiol. 41, 628–634.

Hinsby, K., McKay, L. D., Jorgensen, P., Lenczewski, M., and Gerba C. P. 1996. Fracture aperturemeasurements and migration of solutes, viruses and immiscible creosote in a column of clay-rich till. Ground Water 34, 1065–1075.

Hurst, C. J., Gerba, C. P., and Cech, I. 1980. Effects of environmental variables and soils character-istics on virus survival in soil. Appl. Environ. Microbiol. 40, 1067–1079.

Hurst, C. J., Wild, D. K., and Clark, R. M. 1992. Comparing the accuracy of equation formats formodelling microbial population decay rates. In: Modelling the Metabolic and PhysiologicActivities of Micro-Organisms, pp. 149–175. (Hurst, C. J., Ed.) New York, John Wiley & Sons.

Jansons, J., Edmonds, L. W., Speight, B., and Bucens, M. R. 1989a. Movement of viruses afterartificial recharge. Water Res. 23, 293–299.

Jansons, J., Edmonds, L. W., Speight, B., and Bucens, M. R.1989b. Survival of viruses in ground-water. Water Res. 23, 301–306.

Jin, Y., Yates, M. V., Thompson, S. S., and Jury, W. A. 1997. Sorption of viruses during flow throughsaturated sand columns. Environ. Sci. Technol. 31, 548–555.

Keswick, B. H. and Gerba, C. P. 1980. Viruses in groundwater. Environ. Sci. Technol. 14, 1290–1297.



124

Keswick, B. H., Gerba, C. P., Secor, S. L., and Cech, I. 1982. Survival of enteric viruses and indicatorbacteria in groundwater. J. Environ. Sci. Health A12, 903–912.

Kinoshita, T., Bales, R. C., Maguire, K. M., and Gerba, C. P. 1993. Effect of pH on bacteriophagetransport through sandy soils. J. Contam. Hydrol. 14, 55–70.

Knorr, M. 1937. Das Gas- und Wasserfach 80 (21), 330–334.Lance, J. C., Gerba, C. P., and Wang, D. S. 1982. Comparative movement of different enteroviruses

in soil columns. J. Environ. Qual. 11, 347–351.Landry, E. F., Vaughn, J. M., Vicalet, T. J., and Mann, R. 1983. Accumulation of sediment-associated

viruses in shellfish. Appl. Environ. Microbiol. 46, 673–682.Levine, A. D., Tchobanoglous, G., and Asano, T. 1985. Characterisation of the size distribution of

contaminants in wastewater: treatment and reuse implications. J. Water Pollut. Control Fed. 57,805–816.

Liew, P. F. and Gerba, C. P. 1980. Thermostabilization of enteroviruses by estuarine environment.Appl. Environ. Microbiol. 40, 305–308.

Lipson, S. M. and Stotzky, G. 1983. Adsorption of reovirus to clay minerals: effects of cation-exchange capacity, cation saturation and surface area. Appl. Environ. Microbiol. 46, 673–682.

Loveland, J. P., Ryan, J. N., Amy, G. L., and Harvey, R. W. 1996 The reversibility of virusattachment to mineral surfaces. Colloids Surfaces A: Physicochem. Eng. Aspects 107, 205–221.

Lytle, C.D. and Routson, L.B. 1995. Minimised virus binding for test barrier materials. Appl.Environ. Microbiol. 61, 643–649.

Macler, B. A. 1996. Developing the Ground Water Disinfection Rule. J. Am. Water Works. Assoc.88, 47–55.

Martin, R. E., Bouwer, E. J. and Hanna, L. M. 1992. Application of clean bed filtration theory tobacterial deposition in porous media. Environ. Sci. Technol. 26, 1053–1058.

Mathess, G., Pekdeger, A., and Schroeter, J. 1988. Persistence and transport of bacteria and virusesin groundwater — a conceptual evaluation. J. Contam. Hydrol. 2, 171–188.

McKay, L. D., Gillham, R. W., and Cherry, J. A. 1993. Field experiments in a fractured clay till. 2.Solute and colloid transport. Water Resource Res. 29, 3879–3890.

Metcalf, T. G., Rao, V. C., and Melnick, J. L. 1984. Solid associated viruses in a polluted estuary.Monogr. Virol. 15, 97–110.

Moore, R. S., Taylor, D. H., and Sturman, L. S. 1981. Poliovirus adsorption by 34 minerals and soils.Appl. Environ. Microbiol. 42, 963–975.

Moore, R. S., Taylor, D. H., and Sturman, L. S. 1982. Adsorption of reovirus by minerals and soils.Appl. Environ. Microbiol. 44, 852–859.

Nasser, A. M., Tchorch, Y., and Fattal, B. 1993. Comparative survival of E. coli, F+bacteriophages,HAV and poliovirus 1 in wastewater and groundwater. Water Sci. Technol. 27, 401–407.

Ouyang, Y., Shinde, D., Mansell, R. S., and Harris, W. 1996. Colloid-enhanced transport of chemi-cals in subsurface environments: a review. Crit. Rev. Environ. Sci. Technol. 26, 189–204.

Park, N. S., Blandford, T. N., and Huyacorn, P. S. 1994. VIRALT— A Model for Simulating ViralTransport in Groundwater. Documentation and User’s Guide. Version 3.0, Herndon, VA,HydroGeologic.

Park, N. S., Blandford, T. N., Wu ,Y. S., and Huyacorn, P. S. 1995. CANVAS- A CompositeAnalytical-Numerical Model for Viral and Solute Transport Simulation. Documentation anduser’s guide. Version 2.1, Herndon, VA, HydroGeologic.

Paul, J. H., Rose, J. B., Brown, J., Shinn, E. A., Miller, S., and Farrah, S. R. 1995. Viral tracer studiesindicate contamination of marine waters by sewage disposal practices in Key Largo, Florida.Appl. Environ. Microbiol. 61, 2230–2234.

Paul, J. H., Rose, J. B., and Jiang, S. C. 1997. Evidence for groundwater and surface marine watercontamination by waste disposal wells in the Florida Keys. Water Res. 31, 1448–1454.

Payment, P., Morin, E., and Trudel, M. 1988. Coliphages and enteric viruses in the particulate phaseof river water. Can. J. Microbiol. 34, 907–910.



125

Penrod, S. L., Olson, T. M., and Grant, S. B. 1995. Whole particle microelectrophoresis for smallviruses. J. Coll. Interface Sci. 173, 521–523.

Penrod, S. L., Olson, T. M., and Grant, S. B. 1996. Deposition kinetics of two viruses in packed bedsof quartz granular media. Langmuir 12, 5576–5587.

Pieper, A. P., Ryan, J. N., Harvey, R. W., Amy, G. L., Illangasekare, T. H., and Metge, D. W. 1997.Transport and recovery of bacteriophage PRD1 in a sand and gravel aquifer: effect of sewage-derived organic matter. Environ. Sci. Technol. 31, 1163–1170.

Poletika, N. N., Jury, W. A., and Yates, M. V. 1995. Transport of bromide, simazine, and MS-2coliphage in a lysimeter containing undisturbed, unsaturated soil. Water Resource Res. 31, 801–810.

Powelson, D. K., Simpson, J. R., and Gerba, C. P. 1990. Virus transport and survival in saturated andunsaturated flow through soil columns. J. Environ. Qual. 19, 396–401.

Powelson, D. K., Simpson, J. R., and Gerba, C. P. 1991. Effects of organic matter on virus transportin unsaturated flow. Appl. Environ. Microbiol. 57, 2192–2196.

Powelson, D. K., Gerba, C. P., and Yahya, M. T. 1993. Virus transport and removal in wastewaterduring aquifer recharge. Water Res. 27, 583–590.

Powelson, D.K. and Gerba, C.P. 1994. Virus removal from sewage effluents during saturated andunsaturated flow through soil columns. Water Res. 28, 2175–2181.

Rao, V. C., Siedel, K. M., Goyal, S. M., Metcalf, T. G., and Melnick, J. L. 1984. Isolation ofenteroviruses from water, suspended solids and sediments from Galveston Bay: Survival ofpoliovirus and rotavirus adsorbed to sediment. Appl. Environ. Microbiol. 48, 404–409.

Redman, J. A., Grant, S. B., Olson, T. M., Hardy, M. E., and Estes, M. K. 1997. Filtration ofrecombinant Norwalk virus particles and bacteriophage MS2 in quartz sand: importance ofelectrostatic interactions. Environ. Sci. Technol. 31, 3378–3383.

Redman, J. A., Grant, S. B., Olson, T. M., Adkins, J. M., Jackson, J. L., Castillo, M. S., and Yanko,W. A. 1999. Physicochemical mechanisms responsible for the filtration and mobilization of afilamentous bacteriophage in quartz sand. Water Res. 33, 43–52.

Regli, S., Rose, J. B., Haas, C. N., and Gerba, C. P. 1991. Modeling the risk from Giardia and virusesin drinking water. J. Am. Water Works Assoc. 213, 76–84.

Rehman, L. L. C., Welty, C., and Harvey, R. W. 1999. Stochastic analysis of virus transport inaquifers. Water Resource Res. 35, 1987–2006.

Rossi, P. 1994. Advances in biological tracer techniques for hydrology and hydrogeology usingbacteriophages. Ph.D. Thesis, University of Neuchatel, Switzerland, 200 pp.

Rossi, P., De Carvalho-Dill, A., Muller, I., and Aragno, M. (1994). Comparative tracing experimentsin a porous aquifer using bacteriophages and fluorescent dye on a test field located at Wilerwald(Switzerland) and simultaneously surveyed on a local scale by radio-magneto-tellury (12–240kHz). Environ. Geol. 23, 192–200.

Ryan, J. N. and Elimelech, M. 1996. Colloid mobilization and transport in groundwater. ColloidsSurfaces A: Physicochem Eng. Aspects 107, 1–56.

Ryan, J. N., Elimelech, M., Ard, R. A., Harvey, R. W., and Johnson, P. R. 1999. Bacteriophage PRD1and silica colloid transport and recovery in an iron oxide-coated sand aquifer. Environ. Sci.Technol. 33, 63–73.

Sakoda, A., Sakai, Y., Hayakawa, K., and Suzuki, M. 1997. Adsorption of viruses in water environ-ment onto solid surfaces. Water Sci. Technol. 35, 107–14.

Schijven, J. F. and Rietveld, L. C. 1996. How do field observations compare with models of microbialremoval? In: The 1996 Proceedings of the Groundwater Foundation’s 12th Annual Fall Sym-posium, pp. 105–114. Boston, MA.

Schijven, J. F., Medema, G. J., de Nijs, A. C. M., and Elzenga, J. G. 1996. Emission and dispersionof Cryptosporidium, Giardia and enteroviruses via domestic wastewater. National Institute ofPublic Health and the Environment, Bilthoven, The Netherlands. Report number 289202014 [inDutch].



126

Schijven, J. F., Hoogenboezem, W., Nobel, P. J., Medema, G. J., and Stakelbeek, A. 1998. Reductionof FRNA-bacteriophages and faecal indicator bacteria by dune infiltration and estimation ofsticking efficiencies. Water Sci. Technol. 38, 127–131.

Schijven, J. F, Hoogenboezem, W., Hassanizadeh, S. M., and Peters, J. H. 1999. Modelling removalof bacteriophages MS2 and PRD1 by dune infiltration at Castricum, the Netherlands. WaterResource Res. 35, 1101–1111.

Shields, P. A. and Farrah, S. R. 1983. Influence of salts on electrostatic interactions betweenpoliovirus and membrane filters. Appl. Environ. Microbiol. 45, 526–531.

Shields, P. A. and Farrah, S. R. 1987. Determination of the electrostatic and hydrophobic characterof enteroviruses and bacteriophages. Abstr. Q-82. Program Abstr. 87th Annu. Meet. Am. Soc.Microbiol. American Society for Microbiology, Washington, D.C.

Simizu, Y., Sogabe, H., and Terashima, Y. 1998. The effects of colloidal humic substances on themovement of non-ionic hydrophobic organic contaminants in groundwater. Water Sci. Technol.38, 159–167.

Sim, Y. and Chrysikopoulos, C. V. 1995. Analytical models for one-dimensional virus transport insaturated porous media. Water Resource Res. 31, 1429–1437.

Sim, Y. and Chrysikopoulos, C. V. 1996. One-dimensional virus transport in porous media with time-dependent inactivation rate coefficients. Water Resource Res. 32, 2607–2611.

Sim, Y. and Chrysikopoulos, C. V. 1998. Three-dimensional analytical models for virus transport insaturated porous media. Transp. Porous Media 30, 87–112.

Singh, S. N., Bassous, M., Gerba, C. P., and Kelley, L. M. 1986. Use of dyes and proteins asindicators of virus adsorption to soils. Water Res. 20, 267–272.

Sinton, L. W., Finlay, R. K., Pang, L., and Scott, D. M. 1997. Transport of bacteria and bacterioph-ages in irrigated effluent into and through an alluvial gravel aquifer. Water Air Soil Pollut. 98,17–42.

Smith, E. M., Gerba, C. P., and Melnick, J. L. 1978. Role of sediment in the persistence ofenteroviruses in the estuarine environment. Appl. Environ. Microbiol. 35, 685–689.

Sobsey, M. D., Dean, C. H., Knuckles, M. E., and Wagner, R. A. 1980. Interaction and survival ofenteric viruses in soil materials. Appl. Environ. Microbiol. 40, 92–101.

Sobsey, M. D., Shields, P. A., Hauchmann, F. H., Hazard, R. L., and Caton, L. W. 1986. Survivaland transport of hepatitis A virus in soils, groundwater, and wastewater. Water Sci. Technol. 18,97–106.

Sobsey, M. D., Hall, R. M., and Hazard, R. L. 1995. Comparative reductions of hepatitis A virus,enteroviruses and coliphage MS2 in miniature soil columns. Water Sci. Technol. 31, 203–209.

Swanton, S.W. 1995. Modelling colloid transport in groundwater; the prediction of colloid stabilityand retention behaviour. Adv. Colloid Interface Sci. 54, 129–208.

Taylor, D. H., Bellamy, A. R., and Wilson, A. T. 1980. Interactions of bacteriophages R17 andreovirus type III with the clay mineral allophane. Water Res. 14, 339–346.

Taylor, D. H., Moore, R. S., and Sturman, L. S. 1981. Influence of pH and electrolyte compositionon adsorption of poliovirus by soils and minerals. Appl. Environ. Microbiol. 42, 976–84.

Thompson, S. S., Flury, M., Yates, M. V., and Jury, W. A. 1998. Role of the air-water-solid interfacein bacteriophage sorption experiments. Appl. Environ. Microbiol. 64, 304–309.

Tim, U. S. and Mostaghimi, S. 1991. Model for predicting virus movement through soils. GroundWater 29, 251–259.

Toride, N., Leij, F. J., and van Genuchten, M. T. 1995. The CXTFIT code for estimating transportparameters from laboratory or field tracer experiments. Version 2.0. U.S. Salinity Laboratory,Agricultural Research Service, U.S. Department of Agriculture, Riverside, CA. Report No. 137.

Vilker, V. L. and Burge, W. D. 1980. Adsorption mass transfer model for virus transport in soils.Water Res. 14, 783–790.

Vilker, V. L., Meronek, G. C., and Bulter, P. C. 1983. Interaction of poliovirus with montmorilloniteclay in phosphate-buffered saline. Eviron. Sci. Technol. 17, 631–634.



127

Wait, D. A. and Sobsey, M. D. 1983. Method for recovery of enteric viruses from estuarine sedimentswith chaotropic agents. Appl. Environ. Microbiol. 46, 379–385.

Wang, D. S., Gerba, C. P., and Lance, J. C. 1981. Effect of soil permeability on virus removal throughsoil. Appl. Environ. Microbiol. 42, 83–88.

Yahya, M. T., Galsomies, L., Gerba, C. P., and Bales, R. C. 1993. Survival of bacteriophages MS2and PRD1 in groundwater. Water Sci. Technol. 27, 409–412.

Yao, K. M., Habibian, M. T., and O’Melia, C. R. 1971. Water and waste water filtration: conceptsand applications. Environ. Sci. Technol. 5, 1105–1112.

Yates, M. V. 1995. Field evaluation of the GWDR’s natural disinfection criteria. J. Am. Water WorksAssoc. 87, 76–85.

Yates, M. V., Gerba, C. P., and Kelley, L. M. 1985. Virus persistence in groundwater. Appl. Environ.Microbiol. 49, 778–781.

Yates, M. V. and Ouyang, Y. 1992.VIRTUS, a model of virus transport in unsaturated soils, Appl.Environ. Microbiol. 58, 1609–1616.

Yates, M. V., Yates, S. R., Warrick, A. W., and Gerba, C. P. 1986. Use of geostatistics to predict virusdecay rates for determination of septic tank setback distances. Appl. Environ. Microbiol. 52,479–483.

Yates, M. V. and Yates, S. R. 1987. A comparison of geostatistical methods for estimating virusinactivation rates in groundwater. Water Res. 21, 1119–1125.

Yates, M. V, Yates, S. R., Wagner J., and Gerba C. P. 1987. Modelling virus survival and transportin the subsurface. J. Contam. Hydrol. 1, 329–345.

Yates, M. V. and Yates, S. R. 1988. Virus survival and transport in groundwater. Water Sci. Technol.20, 301–307.

Documents

Removal of Viruses by Soil Passage: Overview of Modeling ... Microbe Removal/3 Virus Sp… · moved. Virus removal is the disappearance of viruses from the system and is defined in